A method and system for screening salt-tolerant varieties based on fluorescence imaging

Abstract

The invention discloses a method and system for screening salt-tolerant varieties based on fluorescence imaging, and specifically relates to the field of variety screening. The method comprises the following steps: a multifunctional plant photosynthetic phenotype measurement system is used to execute a PAM Time protocol program to completely measure the fluorescence quenching kinetic curve of a plant to be tested; then, the Pearson correlation coefficient is applied to mine the correlation of data; then, the original data is subjected to dimensionality reduction processing by principal component analysis, and the membership function of a comprehensive salt-tolerant coefficient is constructed by combining the principal components extracted by PCA and the index weights, and a D value is calculated; the D value is used as a dependent variable to construct a multivariate stepwise regression model including fluorescence parameters as independent variables; finally, a one-dimensional convolutional neural network based on an AlexNet framework is constructed by using fluorescence data collected by the multifunctional plant photosynthetic phenotype measurement system, and a comprehensive discrimination model of a support vector machine, an extreme learning machine and a partial least squares discriminant analysis is constructed at the same time; and a comprehensive salt-tolerant identification system is established by grading according to the comprehensive discrimination model and system clustering to screen varieties.

Description

Technical Field

[0001] The present invention relates to the technical field of variety screening, and more specifically, to a method and system for screening salt-tolerant varieties based on fluorescence imaging. Background Art

[0002] Excessive salt is one of the key adverse factors that restrict the normal growth and development of plants, which can lead to stunted growth, damaged development and even death of plants. Therefore, screening out salt-tolerant crop varieties in agricultural production is crucial to increasing crop yields.

[0003] The current mainstream salt stress detection methods focus on quantitative evaluation through physiological and biochemical indicators, such as monitoring changes in specific ion concentrations in plants, measuring a series of physiological parameters, and recording plant morphological characteristic data.

[0004] However, in actual use, it still has some shortcomings, such as cumbersome operation, long time consumption, high cost, large manpower demand and low efficiency. In addition, in the analysis stage after obtaining the data, the existing processes and technologies also have a certain lag. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method and system for screening salt-tolerant varieties based on fluorescence imaging to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] Step A1: Use the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Time protocol program to measure the fluorescence of the cabbage, and completely measure the fluorescence quenching kinetic curve of the cabbage to be tested;

[0008] Step A2: applying the Pearson correlation coefficient to represent the correlation between different fluorescence features;

[0009] Step A3: Reduce the dimension of the original data through principal component analysis and transform it into a set of linearly independent comprehensive indicators;

[0010] Step A4: Combining the principal components extracted by principal component analysis with the index weights, constructing the membership function of the comprehensive salt tolerance coefficient, and calculating the comprehensive salt tolerance coefficient D value;

[0011] Step A5: Conduct an in-depth screening of the salt tolerance of cabbage, take the D value as the dependent variable, and construct a multivariate stepwise regression model with the fluorescence parameters as the independent variables;

[0012] Step A6: Build based on A one-dimensional convolutional neural network of the framework is used, and a comprehensive discriminant model of support vector machine, extreme learning machine and partial least squares discriminant analysis is constructed at the same time.

[0013] Preferably, in step A1, the steps of using the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Time protocol program to completely measure the fluorescence quenching kinetic curve of the plant to be tested are specifically as follows:

[0014] Step A11: preparing a plurality of varieties of cabbage samples to be tested having the same growth conditions as the control and the cabbage samples to be tested under the treatment of salt with different concentrations;

[0015] Step A12: Start the measurement program, and the system will automatically record the changes of photosynthetic parameters and fluorescence parameters over time;

[0016] Step A13: During the measurement process, the system will automatically record the changes of photosynthetic parameters over time, including fluorescence parameters F0, Fv' / Fm', Fm, Fv / Fm, Fq' / Fm', Fs', Fm', rETR, NPQ, F0', qP, qN, qL, фno, фnpq, npq(t), Red, Green, Blue, SpcGrn, FarRed, Nir, ChlIdx, AriIdx, NDVI, dChl, Chl, aRed, aFarRed, and Alpha.

[0017] The specific steps of fluorescence imaging are as follows:

[0018] The collected fluorescence image is subjected to background correction and noise removal; image segmentation is performed to extract structures or regions of interest; and image analysis software is used to extract data from the fluorescence signal. Preferably, in step A2, the calculation method of the Pearson correlation coefficient is specifically as follows:

[0019] , where C is the Pearson correlation coefficient, It is expressed as the covariance of g and y, where g and y represent different fluorescence parameters. Expressed as the characteristic standard deviation of g, Expressed as the characteristic standard deviation of y.

[0020] When the Pearson correlation coefficient is calculated , it means that there is zero correlation between the photosynthetic rate g and the chlorophyll fluorescence value y, and there is no linear relationship between different fluorescence parameters;

[0021] When the Pearson correlation coefficient is calculated , it means that the different fluorescence parameters are completely correlated, that is, the different fluorescence parameters can be described by a straight line equation, and all the points fall on a straight line;

[0022] When the calculated Pearson correlation coefficient is a positive number, it means that the chlorophyll fluorescence value y increases with the increase of the photosynthetic rate g; when the calculated Pearson correlation coefficient is a negative number, it means that the chlorophyll fluorescence value y decreases with the increase of the photosynthetic rate g.

[0023] Preferably, in step A3, the principal component analysis method is specifically:

[0024] There are n indicator variables in principal component analysis. The value of the bth indicator of the ath evaluation object is , transforming each indicator into a standardized comprehensive indicator ;

[0025] ,in Expressed as a standardized comprehensive index, It is represented by the value of the bth indicator of the ath evaluation object. It is expressed as the average value of all evaluation objects of the bth indicator, Expressed as the sample standard deviation of the b-th indicator.

[0026] Preferably, the calculation method of the membership function is specifically as follows:

[0027] , where L represents the membership function, It is expressed as the score of the cth principal component of the dth accession, is expressed as the minimum value of the cth component, Expressed as the maximum value of the cth component;

[0028] If the mth indicator is negatively correlated with the trait it belongs to, the calculation method of the inverse membership function is as follows:

[0029] ,in, is expressed as the inverse membership function, It is expressed as the score of the cth principal component of the dth accession, is expressed as the minimum value of the cth component, Expressed as the maximum value of the cth component;

[0030] The specific method for calculating the weight is:

[0031] ,in, It is expressed as the weight of the jth principal component among all principal components. It is expressed as the contribution rate of the jth principal component obtained through principal component analysis;

[0032] The calculation method of comprehensive salt tolerance coefficient is:

[0033] , where D represents the comprehensive salt tolerance coefficient, Expressed as a membership function, It is expressed as the weight of the jth principal component among all principal components.

[0034] Preferably, the algorithm of the regression equation is specifically:

[0035] , where D represents the comprehensive salt tolerance coefficient, Expressed as wavelengths between 700 nm and 750 nm Light; It is expressed as the non-photochemical quenching coefficient, dFv / Fm is expressed as the excitation energy capture efficiency of the PSII reaction center opened under light, Expressed as maximum fluorescence, It is represented by the effective photochemical quantum yield of PSII, Red is represented by the intensity of red light, It is expressed as the content of chlorophyll. The salt tolerance of Chinese cabbage seedlings was graded according to D value and system clustering.

[0036] Preferably, in step A6, based on The one-dimensional convolutional neural network of the framework uses distributed training and GPU acceleration to speed up the learning progress for large-scale data processing. The one-dimensional convolutional neural network can be trained as a complete end-to-end system; the support vector machine strives to maximize the classification boundary to enhance the generalization performance of the model, the extreme learning machine focuses on achieving fast training, and the partial least squares discriminant analysis is mainly used to process highly correlated fluorescence indicators; all models adopt a five-fold cross-validation strategy. After five training and five verifications, the five-fold cross-validation accuracy and the average value of its evaluation parameters are finally selected as the overall performance indicator of the model, and the evaluation parameters include F1, Precision, and Recall; based on the above models, a comprehensive salt-tolerance identification system is established to screen varieties.

[0037] Technical effects and advantages of the present invention:

[0038] 1. The present invention adopts a novel plant physiological indicator system, combined with the photosynthesis ability characteristics of the plant itself, to improve the credibility of salt-tolerant variety identification.

[0039] 2. In actual operation, this invention can reduce the time required for data collection and speed up the identification process of salt-tolerant varieties.

[0040] 3. By integrating photosynthetic phenotypic imaging technology, machine learning and convolutional neural network technology, this invention ensures that the accuracy of identification results is greatly improved, thereby providing a more efficient and accurate method for the screening of salt-tolerant crop varieties. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 The present invention is a schematic diagram of a method for screening salt-tolerant varieties based on fluorescence imaging.

[0042] Figure 2 This is a schematic diagram of the module connections of a salt-tolerant variety screening system based on fluorescence imaging of the present invention.

[0043] Figure 3 Schematic diagram of the degree of explanation of each indicator based on the principal component analysis results

[0044] Figure 4 Schematic diagram of the positive and negative correlation between various indicators

[0045] Figure 5 Schematic diagram of recognition accuracy and evaluation parameters of each model DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] See also Figure 1 As shown, the present invention provides a method for screening salt-tolerant varieties based on fluorescence imaging, and the method is:

[0048] Step A1: Use the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Time protocol program to measure the fluorescence of the cabbage, and completely measure the fluorescence quenching kinetic curve of the cabbage to be tested;

[0049] In step A1, the steps of using the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Time protocol program to completely measure the fluorescence quenching kinetic curve of the plant to be tested are specifically as follows:

[0050] Step A11: preparing a Chinese cabbage sample to be tested that is healthy and in good growth condition;

[0051] Step A12: placing the cabbage sample in the multifunctional plant photosynthetic phenotype measurement system, and setting the light intensity, temperature, and humidity;

[0052] Step A13: Select the PAM Time Protocol program in the multifunctional plant photosynthetic phenotype measurement system, and set the measurement time interval and duration according to the system prompts;

[0053] Step A14: Start the measurement program, and the system will automatically record the changes of photosynthetic parameters and fluorescence parameters over time;

[0054] Step A15: During the measurement process, the system automatically records the changes of photosynthetic parameters over time, wherein the photosynthetic parameters include photosynthetic rate and chlorophyll fluorescence; after the measurement, the data is exported to a computer for fluorescence quenching kinetic curve fitting;

[0055] The specific steps of fluorescence imaging are as follows:

[0056] The collected fluorescence images are subjected to background correction and noise removal; image segmentation is performed to extract structures or regions of interest; and image analysis software is used to extract data from the fluorescence signal.

[0057] In step A2, the calculation method of the Pearson correlation coefficient is specifically as follows:

[0058] , where C is the Pearson correlation coefficient, It is expressed as the covariance of g and y, where g and y represent different fluorescence indicators. Expressed as the characteristic standard deviation of g, Expressed as the characteristic standard deviation of y.

[0059] When the Pearson correlation coefficient is calculated , it means that there is zero correlation between different fluorescence indices, and there is no linear relationship between different fluorescence indices;

[0060] When the Pearson correlation coefficient is calculated , it means that the different fluorescence indices are completely correlated, that is, the different fluorescence indices can be described by a straight line equation, and all the points fall on a straight line;

[0061] When the calculated Pearson correlation coefficient is a positive number, it means that the fluorescence index y increases as the fluorescence index g increases; when the calculated Pearson correlation coefficient is a negative number, it means that the fluorescence index y decreases as the fluorescence index g increases.

[0062] Step A3: Reduce the dimension of the original data through principal component analysis and transform it into a set of linearly independent comprehensive indicators;

[0063] In step A3, the principal component analysis method is specifically as follows:

[0064] There are n indicator variables in principal component analysis. The value of the bth indicator of the ath evaluation object is , transforming each indicator into a standardized comprehensive indicator ;

[0065] ,in Expressed as a standardized comprehensive index, It is represented by the value of the bth indicator of the ath evaluation object. It is expressed as the average value of all evaluation objects of the bth indicator, Expressed as the sample standard deviation of the bth indicator;

[0066] ,in is represented by an indexed indicator variable, n is represented by n indicator variables, It is represented by the value of the bth indicator of the ath evaluation object;

[0067] ,in , represents the sample standard deviation of the bth indicator, n represents the nth indicator variable, It is represented by the value of the bth indicator of the ath evaluation object. Expressed as an index variable.

[0068] Step A4: Combining the principal components extracted by principal component analysis with the index weights, constructing the membership function of the comprehensive salt tolerance coefficient, and calculating the comprehensive salt tolerance coefficient;

[0069] The calculation method of the membership function is as follows:

[0070] , where L represents the membership function, It is expressed as the score of the cth principal component of the dth accession, is expressed as the minimum value of the cth component, Expressed as the maximum value of the cth component;

[0071] If the mth indicator is negatively correlated with the trait it belongs to, the calculation method of the inverse membership function is as follows:

[0072] ,in, is expressed as the inverse membership function, It is expressed as the score of the cth principal component of the dth accession, is expressed as the minimum value of the cth component, Expressed as the maximum value of the cth component;

[0073] The specific method for calculating the weight is:

[0074] ,in, It is expressed as the weight of the jth principal component among all principal components. It is expressed as the contribution rate of the jth principal component obtained through principal component analysis;

[0075] The calculation method of comprehensive salt tolerance coefficient is:

[0076] , where D represents the comprehensive salt tolerance coefficient, Expressed as a membership function, It is expressed as the weight of the jth principal component among all principal components;

[0077] If the calculated comprehensive salt tolerance coefficient is greater than the preset comprehensive salt tolerance coefficient threshold, then the cabbage sample is salt-tolerant; if the calculated comprehensive salt tolerance coefficient is less than the preset comprehensive salt tolerance coefficient threshold, then the cabbage sample is sensitive; if the calculated comprehensive salt tolerance coefficient is equal to the preset comprehensive salt tolerance coefficient threshold, then the cabbage sample is normal; the calculated data are clustered and divided into salt tolerance, sensitivity and normality.

[0078] Step A5: Conduct an in-depth screening of the salt tolerance of cabbage, take the D value as the dependent variable, and construct a multivariate stepwise regression model with the fluorescence parameters as the independent variables;

[0079] The specific algorithm of the regression equation is:

[0080] , where D represents the comprehensive salt tolerance coefficient, Expressed as wavelengths between 700 nm and 750 nm Light; Expressed as the non-photochemical quenching coefficient, It is expressed as the excitation energy capture efficiency of the PSII reaction center opened under light, Expressed as maximum fluorescence, It is represented by the effective photochemical quantum yield of PSII, Red is represented by the intensity of red light, It is expressed as the content of chlorophyll. The salt tolerance of Chinese cabbage seedlings was graded according to D value and system clustering.

[0081] Step A6: Build based on A one-dimensional convolutional neural network of the framework is used, and a comprehensive discriminant model of support vector machine, extreme learning machine and partial least squares discriminant analysis is constructed at the same time.

[0082] In step A6, based on The one-dimensional convolutional neural network of the framework uses distributed training and GPU acceleration to speed up the learning progress for large-scale data processing. The one-dimensional convolutional neural network can be trained as a complete end-to-end system; the support vector machine strives to maximize the classification boundary to enhance the generalization performance of the model, the extreme learning machine focuses on achieving fast training, and the partial least squares discriminant analysis is mainly used to process highly correlated fluorescence indicators; all models adopt a five-fold cross-validation strategy. After five training and five verifications, the five-fold cross-validation accuracy and the average value of its evaluation parameters are finally selected as the overall performance indicator of the model, where the evaluation parameters include F1, Precision, and Recall; according to the above models and system clustering, a comprehensive salt tolerance identification system is established for screening varieties;

[0083] Among them, Precision is expressed as the accuracy, Recall is expressed as the recall rate, and F1 is expressed as the harmonic mean of Precision and Recall.

[0084] Among them, support vector machine is represented as SVM, extreme learning machine is represented as ELM, partial least squares discriminant analysis is represented as PLS, one-dimensional convolutional neural network is represented as 1D-CNN, and the comprehensive salt tolerance coefficient is D value.

[0085] See also Figure 2 As shown, in this embodiment, it should be specifically explained that the present invention provides a salt-tolerant variety screening system based on fluorescence imaging, comprising the following modules:

[0086] Fluorescence quenching kinetic curve module: used to use the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Timeprotocol program to perform fluorescence measurement on cabbage, and completely measure the fluorescence quenching kinetic curve of the cabbage to be tested;

[0087] Pearson correlation coefficient calculation module: used to calculate the Pearson correlation coefficient, and use the calculated Pearson correlation coefficient to represent the correlation between different fluorescence indicators;

[0088] Principal component analysis module: used to reduce the dimension of the original data through principal component analysis and transform it into a set of linearly independent comprehensive indicators;

[0089] Comprehensive salt tolerance coefficient calculation module: used to combine the principal components extracted by principal component analysis and index weights, construct the membership function of the comprehensive salt tolerance coefficient, and calculate the comprehensive salt tolerance coefficient;

[0090] Regression equation module: used for in-depth screening of cabbage salt tolerance, taking D value as the dependent variable and constructing a multivariate stepwise regression model with fluorescence parameters as independent variables

[0091] Comprehensive discrimination model module: build based on A one-dimensional convolutional neural network of the framework is used, and a comprehensive discriminant model of support vector machine, extreme learning machine and partial least squares discriminant analysis is constructed at the same time.

[0092] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for screening salt-tolerant varieties based on fluorescence imaging, characterized in that: include: Step A1: Use the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Time protocol program to measure the fluorescence of the cabbage, and completely measure the fluorescence quenching kinetic curve of the cabbage to be tested; Step A2: Apply the Pearson correlation coefficient to represent the correlation between various fluorescence indicators, in preparation for the PCA stage; Step A3: Reduce the dimension of the original data through principal component analysis and transform it into a set of linearly independent comprehensive indicators; Step A4: Combining the principal components extracted by principal component analysis with the index weights, constructing the membership function of the comprehensive salt tolerance coefficient, and calculating the comprehensive salt tolerance coefficient D value; Step A5: Conduct an in-depth screening of the salt tolerance of cabbage, take the D value as the dependent variable, and construct a multivariate stepwise regression model with the fluorescence parameters as the independent variables; The specific model of the regression equation is: , where D represents the comprehensive salt tolerance coefficient, Expressed as wavelengths between 700 nm and 750 nm Light; Expressed as the non-photochemical quenching coefficient, It is expressed as the excitation energy capture efficiency of the PSII reaction center opened under light, Expressed as maximum fluorescence, It is represented by the effective photochemical quantum yield of PSII, Red is represented by the intensity of red light, It is expressed as the content of chlorophyll; the salt tolerance of Chinese cabbage seedlings was graded according to D value and system clustering; This model can be used to accurately quantify the salt tolerance level of a single sample; Combined with D value and system clustering, salt tolerance was divided into salt tolerance, normal type and sensitive type. Step A6: Build based on A one-dimensional convolutional neural network of the framework is used, and a comprehensive discriminant model of support vector machine, extreme learning machine and partial least squares discriminant analysis is constructed at the same time.

2. The method for screening salt-tolerant varieties based on fluorescence imaging according to claim 1, characterized in that: In step A1, the steps of using the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Time protocol program to completely measure the fluorescence quenching kinetic curve of the plant to be tested are specifically as follows: Step A11: preparing a plurality of varieties of cabbage samples to be tested having the same growth conditions as the control and the cabbage samples to be tested under the treatment of salt with different concentrations; Step A12: Start the measurement program, and the system will automatically record the changes of photosynthetic parameters and fluorescence parameters over time; Step A13: During the measurement process, the system will automatically record the changes of photosynthetic parameters over time, including F0, Fv' / Fm', Fm, Fv / Fm, Fq' / Fm', Fs', Fm', rETR, NPQ, F0', qP, qN, qL, фno, фnpq, npq(t), Red, Green, Blue, SpcGrn, FarRed, Nir, ChlIdx, AriIdx, NDVI, dChl, Chl, aRed, aFarRed, and Alpha.

3. The method for screening salt-tolerant varieties based on fluorescence imaging according to claim 1, characterized in that: The specific steps of fluorescence imaging are as follows: The collected fluorescence images are subjected to background correction and noise removal; image segmentation is performed to extract structures or regions of interest; and image analysis software is used to extract data from the fluorescence signal.

4. The method for screening salt-tolerant varieties based on fluorescence imaging according to claim 1, characterized in that: In step A2, the calculation method of the Pearson correlation coefficient is specifically as follows: , where C is the Pearson correlation coefficient, It is expressed as the covariance of g and y, where g and y represent different fluorescence parameters. Expressed as the characteristic standard deviation of g, Expressed as the characteristic standard deviation of y.

5. The method for screening salt-tolerant varieties based on fluorescence imaging according to claim 1, characterized in that: In step A3, the principal component analysis method is specifically as follows: There are n indicator variables in principal component analysis. The value of the bth indicator of the ath evaluation object is , transforming each indicator into a standardized comprehensive indicator .

6. The method for screening salt-tolerant varieties based on fluorescence imaging according to claim 1, characterized in that: The calculation method of the membership function is as follows: , where L represents the membership function, It is expressed as the comprehensive variable value of the cth principal component of the dth cabbage sample. It is expressed as the minimum value of the cth component of all samples, It is expressed as the maximum value of the cth component of all samples. This step is to perform maximum and minimum standardization on each principal component. The specific method for calculating the weight is: ,in, It is expressed as the weight of the jth principal component among all principal components. It is expressed as the contribution rate of the jth principal component obtained through principal component analysis; The calculation method of comprehensive salt tolerance coefficient is: , where D represents the comprehensive salt tolerance coefficient, Expressed as a membership function, It is expressed as the weight of the jth principal component among all principal components.

7. The method for screening salt-tolerant varieties based on fluorescence imaging according to claim 1, characterized in that: In step A6, based on The one-dimensional convolutional neural network of the framework uses distributed training and GPU acceleration to speed up the learning progress for large-scale data processing. The one-dimensional convolutional neural network can be trained as a complete end-to-end system. Among the contrasting algorithms, the support vector machine strives to maximize the classification boundary to enhance the generalization performance of the model, the extreme learning machine focuses on achieving fast training, and the partial least squares discriminant analysis is used to process highly correlated fluorescence indicators.

8. The method for screening salt-tolerant varieties based on fluorescence imaging according to claim 7, characterized in that: All models adopted a five-fold cross-validation strategy. After five trainings and five validations, the average value of the five-fold cross-validation accuracy and its evaluation parameters were finally selected as the overall performance indicator of the model, among which the evaluation parameters included F1, Precision, and Recall. A comprehensive salt-tolerance identification system was established based on the above model level to screen varieties. Among them, Precision is expressed as the accuracy rate, Recall is expressed as the recall rate, and F1 is expressed as the harmonic mean of Precision and Recall.

9. A salt-tolerant variety screening system using fluorescence imaging, using a salt-tolerant variety screening method using fluorescence imaging as described in any one of claims 1 to 8, characterized in that: Includes the following modules: Fluorescence quenching kinetic curve module: used to use the multifunctional plant photosynthetic phenotype measurement system to execute the PAM Timeprotocol program to perform fluorescence measurement on cabbage, and completely measure the fluorescence quenching kinetic curve of the cabbage to be tested; Pearson correlation coefficient calculation module: used to calculate the Pearson correlation coefficient, and use the calculated Pearson correlation coefficient to represent the correlation between various fluorescence indicators; Principal component analysis module: used to reduce the dimension of the original data through principal component analysis and transform it into a set of linearly independent comprehensive indicators; Comprehensive salt tolerance coefficient calculation module: used to combine the principal components extracted by principal component analysis and index weights, construct the membership function of the comprehensive salt tolerance coefficient, and calculate the comprehensive salt tolerance coefficient; Regression equation module: used for in-depth screening of cabbage salt tolerance, taking D value as the dependent variable, constructing a multivariate stepwise regression model with fluorescence parameters as independent variables, and combining D value with system clustering to achieve salt tolerance level classification; Comprehensive discrimination model module: build based on A one-dimensional convolutional neural network of the framework is used, and a comprehensive discriminant model of support vector machine, extreme learning machine and partial least squares discriminant analysis is constructed at the same time.

Images