Method and device for predicting germination of sweet corn seeds

Abstract

This invention relates to the field of crop seed growth prediction technology, and discloses a method for predicting sweet corn seed germination, comprising: acquiring hyperspectral test data of the seeds to be tested; preprocessing the hyperspectral test data to obtain hyperspectral feature data; performing first-order difference model on the hyperspectral feature data to obtain key spectral features related to seed vigor; and calculating based on the key band features and regression analysis algorithm to determine the seed vigor status. This invention also discloses a sweet corn seed germination prediction model. In this embodiment, a correlation is established between seed hyperspectral data and the root length of the seedling after germination, then key spectral information is obtained through hyperspectral feature difference processing, and then regression analysis is used to predict the seedling root length; this enables rapid, non-destructive, and high-precision seed vigor detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crop seed growth prediction, in particular to a sweet corn seed germination prediction method and device. BACKGROUND

[0002] At present, sweet corn (Zea mays L. saccharata) is a kind of vegetable crop with high nutritional and edible value, because it is rich in sugar, various amino acids, vitamins, minerals and dietary fiber. There are many varieties of sweet corn, and it is more popular among consumers all over the world than ordinary corn. It is introduced that in recent years, the planting area of sweet corn in China has gradually expanded. With the continuous improvement of people's requirements for safe production and variety reliability, high-quality seeds have become the top priority for the development of planting industry. However, the vegetation conditions (soil moisture, temperature, nutrition, pests and diseases) of seed planting, the harvesting conditions (mechanical damage, maturity) and the environmental conditions after harvesting (seed drying and storage) are difficult to control. These factors may cause serious loss and slow development of seed growth. Seed vigor test is the focus of modern seed science and the premise of high yield. Seed vigor is an important indicator for measuring the comprehensive germination rate, emergence rate, seedling growth potential and plant stress resistance. Therefore, the seed vigor must be understood before sowing to ensure high seed germination rate and economic benefits. Therefore, it is of important biological and economic significance to establish a rapid, non-destructive and high-precision seed vigor detection method to ensure seed quality, optimize crop production facilities and improve crop yield. SUMMARY

[0003] In view of the defects, the embodiment of the present application discloses a sweet corn seed germination prediction method, which adopts hyperspectral data to predict seedling root length, so as to realize better seed germination prediction and better non-destructive detection.

[0004] The first aspect of the embodiment of the present application discloses a sweet corn seed germination prediction method, comprising:

[0005] A sweet corn seed germination prediction method, comprising:

[0006] Obtaining hyperspectral test data of the seed to be tested;

[0007] Pretreating the hyperspectral test data to obtain seed hyperspectral feature data;

[0008] Using a first-order difference model to obtain features of the hyperspectral data to determine corresponding key band features;

[0009] According to the key band features and a regression analysis algorithm, the seed vigor state is calculated.

[0010] As an optional implementation, in the first aspect of the embodiment of the present application, the pre-processing of the hyperspectral test data to obtain hyperspectral feature data comprises:

[0011] The region of interest of the hyperspectral test data is extracted by using an elliptical segmentation method to obtain interest feature data;

[0012] The interest feature data is subjected to black and white correction by using a black and white correction formula to obtain pre-processed hyperspectral feature data, and the black and white correction formula is:

[0013] ;

[0014] wherein I is the hyperspectral feature data, I raw is the collected seed hyperspectral data, I black is the dark light correction data obtained when the scanning lens is blocked, I white is the white light correction data obtained by scanning a calibration white correction board with a reflectivity of 99.99%.

[0015] As an optional implementation, in the first aspect of the embodiment of the present application, the hyperspectral test data is hyperspectral data in a spectral range of 400nm-1000nm; and the hyperspectral test data is one of hyperspectral data of a seed embryo surface and hyperspectral data of a seed endosperm surface.

[0016] As an optional implementation, in the first aspect of the embodiment of the present application, the feature extraction of the hyperspectral feature data by using a spectral first-order difference information acquisition model to determine corresponding key spectral features comprises:

[0017] The hyperspectral feature data is subjected to feature band division according to a preset number of spectral bands;

[0018] The original spectral data and its corresponding first-order difference features are constructed to determine the correlation between the spectral features and the seedling root length; and the correlation coefficient model is:

[0019] Suppose the original data sequence is y i (1≤ i ≤ n ), the variable y is only defined on non-negative integer values; when the variable i takes non-negative integers in turn, and when i changes from k to k +1, the change amount of the variable is:

[0020] △y k = y k+1 - y k (2)

[0021] △ y k The first-order difference of a function at a point k ;

[0022] The spectral feature information is determined based on the spectral data and the first-order difference information thereof.

[0023] As an optional implementation, in the first aspect of the embodiment of the present application, the number of the preset original spectral band data variables is 220.

[0024] As an optional implementation, in the first aspect of the embodiment of the present application, the calculation according to the key band features and the regression analysis algorithm to determine the seed vigor state comprises:

[0025] The principal component regression, the partial least squares regression and the support vector regression algorithm are used to perform regression analysis on the key feature bands to determine the seed vigor state, and the seed vigor state is the seedling root length.

[0026] As an optional implementation, in the first aspect of the embodiment of the present application, the kernel function selected by the support vector regression algorithm is a Gaussian radial basis function; and the Gaussian radial basis function is:

[0027] wherein σ is selected as σ=50.

[0028] The second aspect of the embodiment of the present application discloses a sweet corn seed germination prediction device, which comprises:

[0029] An acquisition module is configured to acquire hyperspectral test data of seeds to be tested.

[0030] A preprocessing module is configured to preprocess the hyperspectral test data to obtain hyperspectral feature data.

[0031] A band selection module is configured to perform feature selection and feature processing on the hyperspectral feature data by using a spectral first-order difference algorithm to determine corresponding key spectral features.

[0032] A state determination module is configured to calculate a seed vigor state according to the key band features and a regression analysis algorithm.

[0033] The third aspect of the embodiment of the present application discloses an electronic device, comprising: a memory storing executable program codes; a processor coupled with the memory; the processor invokes the executable program codes stored in the memory, and is used for executing the sweet corn seed germination prediction method disclosed in the first aspect of the embodiment of the present application.

[0034] The fourth aspect of the embodiment of the present application discloses a computer readable storage medium storing a computer program, wherein the computer program enables a computer to execute the sweet corn seed germination prediction method disclosed in the first aspect of the embodiment of the present application.

[0035] Compared with the prior art, the embodiment of the present application has the following beneficial effects:

[0036] In the embodiment of the present application, the correlation between the hyperspectral data and the seedling root length after germination is established, then the first-order difference model is used to select the key spectral information of the corresponding hyperspectral data, and then the regression analysis is used to determine the seedling root length prediction; the seed vitality detection can be realized quickly, non-destructively and with high precision. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0038] Figure 1 is a flowchart of the sweet corn seed germination prediction method disclosed in the embodiment of the present application;

[0039] Figure 2 is a flowchart of the preprocessing step disclosed in the embodiment of the present application;

[0040] Figure 3 is a flowchart of the wave band selection disclosed in the embodiment of the present application;

[0041] Figure 4 is a flowchart of the construction of the prediction model disclosed in the embodiment of the present application;

[0042] Figure 5 is a relationship diagram of the correlation coefficient between the hyperspectral data and the seedling root length disclosed in the embodiment of the present application;

[0043] Figure 6 is a regression prediction result diagram of the seedling root length disclosed in the embodiment of the present application;

[0044] Figure 7 is a structural schematic diagram of a sweet corn seed germination prediction device provided by the embodiment of the present application;

[0045] Figure 8 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0047] It should be noted that the terms "first", "second", "third", "fourth" and the like in the specification and claims of the present application are used to distinguish different objects, rather than to describe a specific sequence. The terms "include" and "have" and any variations thereof in the embodiments of the present application are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units need not be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0048] Traditional seed vigor evaluation methods include immunoassay tests, polymerase chain reaction tests, and germination tests. However, the above chemical methods or planting methods are expensive, time-consuming, and destructive, and usually require many instruments. Therefore, they are not suitable for use when a quick and easy estimate of seed vigor is required. In order to achieve accurate monitoring and quality control, a reliable non-destructive testing method is needed. Recently, some non-destructive testing methods have been proposed, such as X-ray diffraction, laser speckle analysis techniques, and electrical conductivity measurements. Due to low efficiency and complex operation, seed detection applications based on these methods are limited. Fortunately, recent studies have shown that spectroscopy-based techniques, such as near-infrared spectroscopy, nuclear magnetic resonance spectroscopy, photoacoustic spectroscopy, hyperspectral, multispectral, and Fourier transform near-infrared spectroscopy, have been developed and successfully applied. In particular, hyperspectral imaging (HSI) is a new technology that combines traditional images with spectral techniques, recording both spectral and spatial information of the object under study, which is very important for seed detection. Therefore, compared with other point spectral techniques that cannot provide spatial information, this method shows great potential in seed vigor evaluation. Therefore, HSI has successfully identified four different varieties of cotton seeds, green coffee beans, hybrid seeds (okra and luffa), hybrid seeds, sweet corn seeds, and waxy corn seeds. In terms of vigor estimation, HSI technology has also been widely used in corn seeds. Taking a short-wave infrared hyperspectral camera in the range of 1000-2500 nm as an example, it was analyzed. The results showed that it is feasible to combine visible near-infrared hyperspectral imaging technology with multiplicative scatter correction (MSC), genetic algorithm (GA), and partial least squares regression (PLSR) and a reliable method for determining the electrical conductivity of corn seeds. In order to detect the vigor of corn seeds during storage, related technical personnel proposed a method for identifying seed vigor under 8 different aging times using HSI, and the results showed the feasibility and effectiveness of using HSI to evaluate seed vigor and seeding degree. Researchers have also developed a new technology for detecting the degree of seed viability affected by microwave heat treatment of three different varieties (yellow, white), and compared the effects of different spectral preprocessing methods, corn seed characteristics, and spectral range on seed germination rate prediction. The accuracy rate of identifying corn seeds (heat-treated) and normal (untreated) seeds using partial least squares discriminant analysis (PLS-DA) was 95.6%. In order to combine the spectral and image information of HSI for seed vigor prediction, a multi-channel data acquisition system was used for image and spectral measurement. High-spectral information was collected from 4 different levels of seed vigor 10 h before germination, and convolutional neural networks were used for vigor evaluation. By comparing different preprocessing and pattern recognition models, the convolutional neural network model integrated spectral and image information and had the best recognition effect, with high prediction accuracy for four levels of vigor. The moisture content, which directly affects the storage time and seed germination rate, was predicted. Based on the hyperspectral images of the two sides (embryo side and endosperm side) of each seed of 4 corn varieties, PLSR prediction was performed.Based on this, the embodiment of the present application discloses a sweet corn seed germination prediction method, device, electronic equipment and storage medium, which establishes the correlation between the hyperspectral data and the root length after germination, then selects the corresponding key band of the hyperspectral by using band selection, and then determines the seedling root length prediction by using regression analysis; which can realize fast, non-destructive and high-precision seed vigor detection.

[0049] Embodiment one

[0050] Please refer to Figure 1 , Figure 1 is the flowchart of the sweet corn seed germination prediction method disclosed by the embodiment of the present application. The execution subject of the method described in the embodiment of the present application is composed of software or / and hardware, which can receive relevant information through wired or / and wireless mode, and can send certain instructions. Of course, it can also have certain processing function and storage function. The execution subject can control multiple devices, such as remote physical servers or cloud servers and related software, or it can be a local host or server and related software that performs related operations on a device placed somewhere. In some scenarios, it can also control multiple storage devices, which can be placed in the same place or different places. As shown in Figure 1 The sweet corn seed germination prediction method includes the following steps:

[0051] S101: Obtain the hyperspectral test data of the seed to be tested;

[0052] More preferably, the hyperspectral test data is the hyperspectral data in the spectral range of 400nm-1000nm; the hyperspectral test data is one of the hyperspectral data of the seed embryo surface or the hyperspectral data of the seed endosperm surface.

[0053] In the specific implementation, a visible / near-infrared hyperspectral imaging system is used, which has a 386.7-1016.7nm band, and includes a CCD camera with a spectral resolution of 3nm±0.5nm, two 50W LED illuminating lamps, and an ND conveyor belt driven by a stepping motor. The system is controlled by a computer with SpVIEW software. Considering the non-uniformity of corn seeds, the embryo surface and endosperm surface of each seed are detected respectively. The above is the specific hyperspectral image acquisition step, but when training, Kennard-Stone method can be used to divide the spectral data of each variety into 73 training sets and 16 test sets. Then the test set is used to calculate the model classification accuracy. Each seed is placed on the workbench with the embryo surface upward and the endosperm surface upward, and then arranged in the same way (8 rows x 8 columns), then scanned at 1.2mm / s, the exposure time is 15ms, and 256 band spectral data is obtained.

[0054] The root length of seedlings after germination reflects seed vigor. This invention utilizes this indicator to monitor seed vigor. Seeds were soaked in water under identical conditions, and germination was continuously observed using the paper tower germination method. Seven days after germination, the root length of the seedling was measured, using the longest root as the standard. The model was then adjusted and trained using these specific data.

[0055] S102: Preprocess the hyperspectral test data to obtain hyperspectral feature data;

[0056] Figure 2 This is a flowchart illustrating the preprocessing steps disclosed in an embodiment of the present invention, as shown below. Figure 2 As shown, more preferably, the preprocessing of the hyperspectral test data to obtain hyperspectral feature data includes:

[0057] S1021: The hyperspectral test data is segmented using an elliptical segmentation method to extract the region of interest to obtain the feature data of interest;

[0058] S1022: The black-and-white correction formula is used to perform black-and-white correction on the feature data of interest to obtain preprocessed hyperspectral feature data. The black-and-white correction formula is:

[0059] ;

[0060] Where I represents hyperspectral feature data. I raw To collect seed hyperspectral data, I black This is low-light correction data obtained when the scanning lens is obstructed. I white The data is white light correction data obtained by scanning a calibration white correction plate with a reflectivity of 99.99%.

[0061] When acquiring hyperspectral images of sweet corn seeds, the spectrometer signal contains not only useful information but also random errors, which can easily cause noise interference in the spectral data. Therefore, 220 bands in the middle of the 430.1nm to 971.5nm band were selected for analysis. Next, regions of interest (ROIs) were created by segmenting the embryo and endosperm of the seed using elliptical divisions, and then the average spectral data within the ROIs were extracted using ENVI 5.1. To ensure that the acquired spectral data reflects the actual spectral characteristics of the seeds, white light correction and dark light correction were performed on the acquired images to eliminate noise effects. Samples were acquired under the same equipment parameter settings, and white light corrected data (I) was obtained by scanning a standard white correction plate with a reflectance of 99.99%. white And cover the scanning lens to obtain low-light correction data I blackUsing two reference values ​​to compare the original spectral image (I) raw Calibration is performed: Based on the above consensus, the corrected hyperspectral data I is calculated.

[0062] S103: Use a first-order difference model of spectroscopy to perform feature selection on the hyperspectral feature data to determine the corresponding key band features;

[0063] Figure 3 This is a schematic diagram illustrating the process of processing and fusing spectral data and its first-order difference feature data as disclosed in the embodiments of the present invention, as follows: Figure 3 As shown, the step of using a spectral feature synthesis model to perform feature fusion on the hyperspectral feature data to determine the corresponding comprehensive hyperspectral feature data includes:

[0064] S1031: Divide the hyperspectral feature data into feature bands according to a preset number of spectral bands;

[0065] S1032: Construct the first-order difference feature of each spectral band data and its corresponding feature to determine the correlation between spectral features and seedling root length; the correlation coefficient model is as follows:

[0066] Assume the original data sequence is y i (1≤ i ≤ n ),variable y Defined only for non-negative integer values; when the variable i Take non-negative integers in sequence, and when i from k Change to k When +1, the change in the dependent variable is:

[0067] △ y k = y k+1 - y k

[0068] △ y k The function is called the function at a point. k The first-order difference;

[0069] Spectral feature information is determined based on spectral data and its first-order difference information;

[0070] S1033: Numerical fusion of the obtained differential information and the original spectral information is performed to obtain complete spectral features.

[0071] More preferably, the number of original spectral bands is 220.

[0072] Each extracted spectrum consists of 220 original spectral bands and 219 first-order difference variables, ranging from 430.1 nm to 971.5 nm. Since fewer variables reduce redundancy and computational cost, it is desirable for fewer frequency bands to represent most of the useful information. To obtain the feature bands most relevant to seed vigor, this embodiment of the invention constructs seed hyperspectral band data and its first-order difference information to obtain data features with a high degree of correlation with seedling root length. Through a regression model, a regression correlation between key seed feature bands and seedling root length is finally established.

[0073] In addition to the methods mentioned above, commonly used hyperspectral data preprocessing methods, such as Savitzky-Golay Smoothing (SGS), Multiplicative ScatterCorrection (MSC), and Standard Normal Variate (SNV), can be used to preprocess the data to obtain the optimal hyperspectral data processing model.

[0074] S104: Calculate the seed vigor status based on the comprehensive spectral characteristics and regression analysis algorithm.

[0075] More preferably, the step of determining seed vigor status based on the spectral characteristics and regression analysis algorithm includes:

[0076] Support vector regression algorithm is used to perform regression analysis on key feature bands to determine seed vigor status, wherein the seed vigor status is seedling root length.

[0077] More preferably, the kernel function selected by the support vector regression algorithm is a Gaussian radial basis function; the Gaussian radial basis function is:

[0078] , where σ is chosen to be σ=50.

[0079] In this embodiment of the invention, principal component regression (PCR), partial least squares (PLS), and support vector regression (SVR) were used to establish a regression relationship between seed hyperspectral data and seedling root length after germination. Hyperspectral analysis provides a comprehensive and complete picture of the seed's external morphology and internal structure. This regression model yielded the relationship between seed hyperspectral data and the corresponding seedling root length, which characterizes seed germination, providing a theoretical basis for non-destructive testing of seed vigor.

[0080] The training data for corn seeds is (D represents 256 hyperspectral wavelengths). The flowchart for model construction and germination prediction is as follows: Figure 4As shown. In practical implementation, various regression algorithms can be used. For example, PCR is a multivariate regression analysis method designed to solve the problem of multicollinearity among independent variables in data regression. In PCR, PCA compresses the collected training data into a low-dimensional space, and then establishes a linear regression model between the projection and output variables to represent the data relationship between the original data and the output variables. The PLS algorithm projects the predictor and observed variables into a new space through data projection, so that the projected data represents the original data as well as best as possible, while the observed variables have a strong explanatory power for the predictor variables. The SVR algorithm aims to obtain an optimal model by obtaining a regression plane, setting the linear function tolerance bias and slack variables, minimizing the loss and minimizing the interval width, in order to obtain the best feature fit between the independent and latent variables. The kernel function chosen by SVR is the Gaussian radial basis function, i.e. , where σ is chosen to be σ=50

[36] . In this embodiment of the invention, three methods are used to perform data regression on the hyperspectral characteristic data of seeds and their corresponding seedling root lengths.

[0081] In this embodiment of the invention, due to the large amount of seed spectral data variables collected, many spectral variables exhibit collinearity, redundancy, and even noise and interference, resulting in excessively long computation time for spectral data regression modeling. This embodiment of the invention employs three types of spectral data feature information: raw hyperspectral data, first-order difference information fused from hyperspectral data, and first-order difference information from hyperspectral data, and uses these spectral feature variables to establish a regression model. Seedling root length is an effective indicator reflecting seed vigor. This embodiment of the invention uses seed spectral data and its first-order difference information as a basis to screen out spectral feature information related to seed vigor. Figure 5 The figure shows the correlation coefficients between the information of each characteristic band and its first-order difference band and the seedling root length. From Figure 5 It can be seen that hyperspectral data is correlated with seedling root length. More importantly, the correlation coefficients between different spectral bands and seedling root length differ, and the correlation between spectral data and their first-order difference with seedling root length shows different characteristics, which can be complementary. From... Figure 5 It can be seen that the high-frequency data of the original spectral data is strongly correlated with the root length of the seedlings, while the mid-to-low frequency data of the first-order difference information is strongly correlated with the root length of the seedlings.

[0082] In the specific implementation, models based on seed hyperspectral data and seedling root length were established using methods such as SVR, PCR, and PLS. The seed vigor prediction performance under different data characteristics, including raw spectral data, spectral data fused with its first-order difference data, and spectral first-order difference data, was compared, and the results were compared with those of different spectral preprocessing methods such as MSC, SGS, and SNV. Table 1 shows the regression and prediction results of seedling root length based on different preprocessing and regression methods, including the RMSE and correlation coefficient R² criteria. A higher correlation coefficient and a lower RMSE indicate higher prediction accuracy and a stronger correlation between the actual and predicted values.

[0083] As shown in Table 1, the seedling root length prediction based on hyperspectral information and first-order difference information outperforms other models in both evaluation criteria. It is noteworthy that the SVR method's model regression is consistent with the actual output trajectory, exhibiting minimal bias. The kernel expansion model outperforms the linear models PCR and PLS, indicating that considering the kernel trick can improve the performance of nonlinear regression and enhance the accuracy of predictions, given that data relationships in seeds are typically nonlinear. Nonlinear models (such as SVR) demonstrate better performance than linear models based on hyperspectral methods. Among different spectral feature extraction methods, the proposed method based on spectral data and its first-order difference data achieved the highest data correlation among all three regression models. By considering the first-order difference of the spectrum in the spectral information, the effectiveness of using hyperspectral first-order difference data for vigor-related modeling is further demonstrated. Table 1 shows that the optimal correlation coefficient is 0.8319, indicating the correlation between hyperspectral data and seed root length. This result further validates the feasibility of hyperspectral methods in seed germination prediction.

[0084] As shown in Table 1, the seedling root length prediction results obtained using SVR, PLS, and PCR regression algorithms based on the fusion of spectral data and its difference data all exhibit the best performance, exceeding the prediction performance of models based on the original spectral data and those based on first-order difference data. In the PLS algorithm, the prediction accuracy obtained based on the original spectral information, spectral difference information, and their fusion features is similar. This may be because the PLS algorithm mainly relies on data correlation analysis, and the variable correlation of the original spectral information is similar to the correlation of its difference information, thus yielding similar results. This is because the data correlation between hyperspectral information and its first-order difference information and seedling root length is complementary, which is also... Figure 5 As shown, their information fusion can better reflect the spectral correlation of seedling root length.

[0085] Table 1 shows that the regression performance of spectral data obtained by commonly used spectral data preprocessing methods such as MSC, SGS, and SNV is poor. This is because commonly used preprocessing methods distort the spectral data, destroying the original spatial distribution characteristics of the spectral data, resulting in a damaged correlation between seed spectral values ​​and seed germination characteristics, affecting the predictive performance of seed vigor characteristics, and leading to larger prediction errors. Table 1 also shows that hyperspectral data of maize seeds have a strong correlation with the root length characteristics of seedlings after germination. This verifies the effectiveness and feasibility of hyperspectral technology in seed germination characteristic prediction, high-quality seed identification, and improved variety breeding. Furthermore, data correlation analysis verifies the effectiveness of germination characteristic prediction based on first-order spectral difference information.

[0086]

[0087] Table 1

[0088] To more clearly study the predictive performance of three regression methods based on first-order difference information, Figure 6 The graph shows the predicted root length of seedlings based on PCR, SVR, and PLS models. It can be noted that all methods were able to capture the data changes in seedling root length because the predicted data matched the actual values ​​for most samples. In particular, the SVR model showed the best agreement between the predicted and actual values, with the smallest deviation.

[0089] The regression results from sweet corn seeds show a strong correlation between hyperspectral data and root length after germination. This confirms the effectiveness of hyperspectral data in seed vigor prediction, high-quality seed identification, and superior variety breeding. Furthermore, the effectiveness of band selection was verified through data correlation. The results indicate that the seedling root length prediction model obtained using the SVR model performs best. SVR is a kernel-based nonlinear regression method that demonstrates a nonlinear relationship between the distribution of seed hyperspectral data and seedling root length. Therefore, the kernel-based nonlinear regression model is the most suitable model for detecting corn seed vigor, achieving good predictive results.

[0090] This invention establishes a maize seed vigor prediction model. The model performance was compared for different pretreatment and spectral feature selections. Most existing methods are based on qualitative analysis and classification of maize seed vigor, such as separation of microwave-treated and untreated seeds, seeding degree assessment, diameter and inactivity discrimination, and aging level prediction. Quantitative analysis mainly focuses on predicting seed vigor-related components. Currently, there are few quantitative studies on maize seed vigor, especially for sweet maize seeds, both domestically and internationally. Sweet maize seeds have insufficient active starch accumulation, high sugar content, low seed vigor, low emergence rate, weak seedling growth, and are susceptible to pathogen infection. Vigor detection methods are mainly focused on common maize seeds with consistent traits, offering high accuracy. However, current methods are not applicable to sweet maize because there are significant differences between varieties after drying. Current seed vigor determination mainly considers field emergence rate, neglecting the effectiveness of seed germination (such as root length and seedling root length). Seed emergence is a crucial foundation for ensuring later crop growth and development. To ensure the systematic and complete evaluation of vigor, evaluating the vigor and emergence ability of sweet maize seeds is particularly important.

[0091] Feature extraction from hyperspectral data is particularly important for sweet corn vigor prediction and variety classification. Although various spectral feature processing and evaluation methods have been proposed, existing hyperspectral preprocessing methods, such as genetic algorithms (GA) and continuous projection algorithms (SPA), are based on global data distribution, primarily focusing on the spatial distribution characteristics of spectral data while neglecting the correlational distribution structure of spectral data variables, such as the correlation between bands and spectral data, and the correlation between spectral data across different bands. Based on this, the applicant proposes a method for expressing sample spectral information based on first-order difference information of spectral data and applies it to seed vigor phenotype prediction research. Utilizing the feature correlation between hyperspectral data and seed vigor phenotypes, three different regression models are constructed based on different spectral data distributions for maize seed vigor prediction research. To evaluate the model accuracy, germination tests are used to measure seedling root length as a reference. The prediction accuracy and superiority of different spectral data acquisition methods and regression algorithms are compared, confirming the effectiveness and feasibility of non-destructive seed vigor detection.

[0092] The embodiments of this invention use hyperspectral data to predict the seed vigor of sweet corn based on PCR, SVR, and PLS models. The main conclusions are as follows.

[0093] (1) Based on regression models such as PCR, SVR, and PLS, a regression correlation model between maize seedling root length and spectral data was established using spectral data and its first difference, ultimately achieving quantitative prediction of seedling root length variable data. The main conclusions are as follows:

[0094] (1) By fusing maize seed spectral data with its first-order difference information, the accuracy of PCR, SVR and PLS algorithms in predicting the root length of seedlings has been improved. Compared with preprocessed data such as MSC, SGS and SNV, as well as curve fitting model data, more accurate prediction results can be obtained by fusing first-order difference spectral information.

[0095] (2) The prediction results show that there are differences in the prediction results under different regression models. The accuracy of the SVR model is higher than that of the PCR and PLS models. The highest determination coefficient of the model is 0.8319, which is more satisfactory than the results of the PCR and PLS models (R2 is 0.8023 and 0.725 respectively).

[0096] (3) The prediction results of the constructed regression model on the seedling root length characteristics show that the actual seedling root length has a high correlation with its corresponding spectral data, reaching a maximum of 0.8319, which verifies the feasibility and effectiveness of seed germination characteristic prediction based on spectral technology. In this embodiment of the invention, a correlation is established between hyperspectral data and root length after germination, then band selection is used to select the corresponding key hyperspectral bands, and then regression analysis is used to determine the seedling root length prediction; it can achieve rapid, non-destructive and high-precision seed vigor detection.

[0097] Example 2

[0098] Please see Figure 7 , Figure 7 This is a schematic diagram of the sweet corn seed germination prediction device disclosed in an embodiment of the present invention. Figure 7 As shown, the sweet corn seed germination prediction device may include:

[0099] Acquisition module 21: Used to acquire hyperspectral test data of the seed to be tested;

[0100] Preprocessing module 22: used to preprocess the hyperspectral test data to obtain hyperspectral feature data;

[0101] Spectral information processing and fusion module 23: used to extract features from the hyperspectral feature data using a first-order spectral difference model to determine key spectral features;

[0102] State determination module 24: used to calculate and determine the seed vigor state based on the key band characteristics and regression analysis algorithm.

[0103] In this embodiment of the invention, a correlation is established between hyperspectral data and root length after germination. Then, first-order difference is used to obtain key spectral information of the hyperspectral data, and regression analysis is used to determine the seedling root length prediction. This enables rapid, non-destructive, and high-precision seed vigor detection.

[0104] Example 3

[0105] Please see Figure 8 , Figure 8 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device can be a computer, a server, etc. Of course, in certain cases, it can also be a mobile phone, tablet computer, monitoring terminal, or other smart device, as well as an image acquisition device with processing capabilities. Figure 8 As shown, the electronic device may include:

[0106] Memory 510 storing executable program code;

[0107] Processor 520 coupled to memory 510;

[0108] The processor 520 calls the executable program code stored in the memory 510 to execute some or all of the steps in the sweet corn seed germination prediction method in Embodiment 1.

[0109] This invention discloses a computer-readable storage medium storing a computer program that causes a computer to perform some or all of the steps in the sweet corn seed germination prediction method of Embodiment 1.

[0110] This invention also discloses a computer program product, wherein when the computer program product is run on a computer, the computer performs some or all of the steps in the sweet corn seed germination prediction method in Embodiment 1.

[0111] This invention also discloses an application publishing platform, which is used to publish computer program products. When the computer program products are run on a computer, the computer executes some or all of the steps in the sweet corn seed germination prediction method in Embodiment 1.

[0112] In various embodiments of the present invention, it should be understood that the sequence number of each process does not necessarily imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0113] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they can be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0114] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0115] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several requests to cause a computer device (which can be a personal computer, server, or network device, specifically a processor in the computer device) to execute some or all of the steps of the methods described in the various embodiments of the present invention.

[0116] In the embodiments provided by this invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0117] Those skilled in the art will understand that some or all of the steps in the various methods of the embodiments described can be implemented by a program instructing related hardware. This program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0118] The foregoing has provided a detailed description of the sweet corn seed germination prediction method, apparatus, electronic device, and storage medium disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for predicting the germination of sweet corn seeds, characterized in that, include: Obtain hyperspectral test data of the seed to be tested; The hyperspectral test data is preprocessed to obtain hyperspectral feature data; The hyperspectral feature data is processed using a first-order spectral difference model to determine the corresponding key band features, including: The hyperspectral feature data is divided into feature bands according to a preset number of spectral bands; Construct the original spectral data and its corresponding first-order difference features to determine the correlation between spectral features and seedling root length; Assume the original data sequence is y i , where 1≤ i ≤ n ,variable y Defined on non-negative integer values; when the variable... i Take non-negative integers in sequence, and when i from k Change to k When +1, the change in the dependent variable is: △ y k = y k+1 - y k △ y k The function is called the function at a point. k The first-order difference; Key spectral features are determined based on spectral data and its first-order difference information. The obtained difference information and the original spectral information are numerically fused to obtain complete spectral features. Each extracted spectrum consists of 220 original spectral bands and 219 first-order difference variables, ranging from 430.1 nm to 971.5 nm. Seed vigor status is determined by calculations based on comprehensive spectral characteristics and regression analysis algorithms. The process of determining seed vigor status based on comprehensive spectral characteristics and regression analysis algorithms includes: Support vector regression algorithm is used to perform regression analysis on key feature bands to determine seed vigor status, and the seedling root length is the state feature reflecting seed vigor status.

2. The method for predicting sweet corn seed germination as described in claim 1, characterized in that, The hyperspectral test data is preprocessed to obtain hyperspectral feature data, including: The hyperspectral test data is segmented using an elliptical segmentation method to extract the region of interest (ROI) to obtain the feature data of interest. The black-and-white correction formula is used to perform black-and-white correction on the feature data of interest to obtain preprocessed hyperspectral feature data. The black-and-white correction formula is as follows: ； Where I represents hyperspectral feature data. I raw To collect seed hyperspectral data, I black This is low-light correction data obtained when the scanning lens is obstructed. I white The data is white light correction data obtained by scanning a calibration white correction plate with a reflectivity of 99.99%.

3. The method for predicting sweet corn seed germination as described in claim 1, characterized in that, The hyperspectral test data is either hyperspectral data of the seed embryo surface or hyperspectral data of the seed endosperm surface.

4. The method for predicting sweet corn seed germination as described in claim 1, characterized in that, The kernel function selected by the support vector regression algorithm is the Gaussian radial basis function.

5. A sweet corn seed germination prediction device, characterized in that, include: Acquisition module: Used to acquire hyperspectral test data of the seed to be tested; Preprocessing module: used to preprocess the hyperspectral test data to obtain hyperspectral feature data; Band selection module: used to select and analyze the hyperspectral feature data using a first-order spectral difference model to determine the corresponding key band features, including: The hyperspectral feature data is divided into feature bands according to a preset number of spectral bands; Construct the original spectral data and its corresponding first-order difference features to determine the correlation between spectral features and seedling root length; Assume the original data sequence is yi, where 1≤i≤n, and the variable... y Defined on non-negative integer values; when the variable... i Take non-negative integers in sequence, and when i from k Change to k When +1, the change in the dependent variable is: △ y k = y k+1 - y k △ y k The function is called the function at a point. k The first-order difference; Key spectral features are determined based on spectral data and its first-order difference information. The obtained difference information and the original spectral information are numerically fused to obtain complete spectral features. Each extracted spectrum consists of 220 original spectral bands and 219 first-order difference variables, ranging from 430.1 nm to 971.5 nm. State determination module: used to calculate and determine seed vigor status based on comprehensive spectral characteristics and regression analysis algorithms; The process of determining seed vigor status based on comprehensive spectral characteristics and regression analysis algorithms includes: Support vector regression algorithm is used to perform regression analysis on key feature bands to determine seed vigor status, and the seedling root length is the state feature reflecting seed vigor status.

6. An electronic device, characterized in that, include: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the sweet corn seed germination prediction method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program causes a computer to perform the sweet corn seed germination prediction method according to any one of claims 1 to 4.

Images