Method for rapidly and nondestructively detecting seed quality based on near infrared spectrum technology

Abstract

The invention discloses a method for rapidly and nondestructively detecting seed quality based on a near infrared spectrum technology, and relates to the technical field of modern agriculture, and the method comprises the following steps: using a deep learning model to carry out image recognition on a seed to be detected, determining a seed type and automatically adjusting spectrometer parameters; collecting spectral data by using the adjusted near-infrared spectrometer, and preprocessing the spectral data; obtaining the internal structure and density information of the seeds, and combining the internal structure and density information with the spectral data to form a multi-modal data set; constructing a seed vigor prediction model, and predicting the germination rate of the seeds; judging the growth trend of the seeds according to the prediction result; grading according to the growth trend of the seeds; according to the method, the seed vigor prediction model is constructed, so that high-precision prediction of the seed germination rate is realized; through classification and grading of growth trends, a scientific basis is provided for seed screening and planting management.

Description

Technical Field

[0001] The invention relates to the field of modern agricultural technology, in particular to a method for rapid and non-destructive seed quality detection based on near infrared spectroscopy technology. Background Art

[0002] In recent years, with the continuous development of modern agricultural technology, seed quality detection has become a key link in improving crop yield and quality. Traditional seed quality detection methods mainly rely on physical and chemical means, such as morphological observation, moisture determination, conductivity measurement, etc. However, these methods are often time-consuming, costly and destructive to seeds, and it is difficult to meet the needs of modern efficient and precision agriculture. With the continuous advancement of optical detection technology, especially the application of near-infrared spectroscopy technology, non-destructive detection of seed quality methods has gradually become a research hotspot. Near-infrared spectroscopy technology can quickly and non-destructively obtain a variety of chemical and physical information of seeds, such as moisture, protein, and fat content, by measuring the absorption spectrum of seeds in the near-infrared region. In addition, combined with image recognition technology and deep learning models, the accuracy and efficiency of detection can be further improved.

[0003] However, the existing near-infrared spectroscopy detection methods still have some shortcomings, which are mainly reflected in the following aspects: First, most methods fail to fully consider the differences in seed types when collecting spectral data, resulting in the fixed spectrometer parameters, affecting the accuracy of the detection results; second, the existing methods usually only rely on a single modality of data for analysis, failing to fully utilize the internal structure and density information of the seeds, limiting the prediction performance of the model; finally, most of the existing classification methods are based on a single indicator for grading, failing to fully consider the growth trend of seeds, resulting in low reliability of the grading results. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for rapid and nondestructive detection of seed quality based on near-infrared spectroscopy technology to solve the problems of fixed parameters, single data and incomplete classification in existing seed quality detection methods.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, an embodiment of the present invention provides a method for rapid and non-destructive detection of seed quality based on near infrared spectroscopy technology, which comprises: Use deep learning models to perform image recognition on seeds to be tested, determine seed type and automatically adjust spectrometer parameters; The spectral data were collected using the adjusted near-infrared spectrometer and preprocessed; Obtain the internal structure and density information of seeds and combine it with spectral data to form a multimodal dataset; Construct a seed vitality prediction model to predict the germination rate of seeds; Determine the growth trend of seeds based on the prediction results; Seeds are graded according to their growth trends.

[0007] As a preferred solution of the method for rapid non-destructive detection of seed quality based on near-infrared spectroscopy technology described in the present invention, the method comprises the following steps: using a deep learning model to perform image recognition on the seeds to be tested, determining the seed type and automatically adjusting the spectrometer parameters. Use a high-resolution digital camera to capture images of seed samples in the sample tray; Grayscale, denoise and edge detect the collected images; Count the number of pixels at each gray level in the image to form a histogram, and calculate the cumulative probability and inter-class variance of each gray level; Traverse all gray levels and select the gray level with the largest inter-class variance as the optimal threshold; Binarize the image according to the optimal threshold; Perform morphological processing on the binarized image to remove small noise points and fill holes; Use convolutional neural network to identify the segmented seed image and determine the seed type; According to the identified seed type, the light source intensity and wavelength range parameters of the spectrometer are automatically adjusted.

[0008] As a preferred embodiment of the method for rapid and non-destructive detection of seed quality based on near-infrared spectroscopy technology described in the present invention, the spectral data includes the moisture content, starch content, fat content and amino acid content of the seeds; and the preprocessing includes moving average, baseline correction, normalization, standard normal variable transformation and detrending processing.

[0009] As a preferred solution of the method for rapid non-destructive detection of seed quality based on near-infrared spectroscopy technology described in the present invention, the steps of obtaining the internal structure image and density information of the seed and combining them with the spectral data to form a multimodal data set include the following steps: Calibrate X-ray imaging equipment and ultrasonic testing equipment using standard samples; Start the calibrated X-ray imaging device to collect images of the internal structure of the seeds; Start the ultrasonic testing equipment to collect the internal density of the seeds; The preprocessed spectral data, internal structure image data and internal density data are numbered respectively, and matched according to the sample numbers; Create a data table containing sample ID, near infrared spectroscopy data, X-ray imaging data, and ultrasonic detection data; Using the sample number as a key, the three data were merged into the same row to form a multimodal dataset and inserted into the data table.

[0010] As a preferred solution of the method for rapid and non-destructive detection of seed quality based on near infrared spectroscopy technology described in the present invention, wherein: constructing a seed vitality prediction model to predict the germination rate of seeds includes the following steps: Select representative seed samples, use high-throughput sequencing technology to sequence the seed genome, and obtain the whole genome sequence; Bioinformatics tools were used to annotate the sequencing data and identify genes associated with germination; RNA was extracted from seeds at different germination stages; High-throughput sequencing technology was used to measure the expression of genes during germination and obtain transcriptome data; Use bioinformatics tools to process high-throughput sequencing data and quantitatively analyze changes in gene expression at different time points; Apply statistical methods to identify genes that are significantly up-regulated and down-regulated during germination. Biomarkers strongly correlated with germination rate were selected from the differentially expressed genes; The selected biomarkers were fused with near-infrared spectroscopy data, X-ray imaging data, and ultrasound detection data to form a comprehensive data set, and a unique number was assigned to each sample; The feature selection algorithm is used to further screen out the features that have the greatest impact on the germination rate and predict the germination rate of seeds. The expression is: ;in, represents the predicted germination rate of seeds, Indicates The comprehensive characteristic value, represents the average value of the comprehensive eigenvector, represents the standard deviation of the comprehensive eigenvector, represents the weight parameter, represents the bias term, Represents the index variable used to identify each comprehensive feature in the comprehensive feature vector.

[0011] As a preferred solution of the method for rapid and non-destructive detection of seed quality based on near infrared spectroscopy technology of the present invention, wherein: judging the growth trend of seeds according to the prediction results includes the following steps: According to the predicted germination rate, the growth trend of seeds is divided into three categories: high growth potential, medium growth potential and low growth potential; The growth threshold of germination rate is set to and ,in Indicates the lower limit of high growth potential, Indicates the lower limit of mesopotamia growth potential; when When When When , it is judged as low growth potential.

[0012] As a preferred solution of the method for rapid and non-destructive seed quality detection based on near infrared spectroscopy technology of the present invention, the classification according to the growth trend of seeds includes the following steps: Read various content data of the test seeds and various content data of the standard seeds from the database as reference standards; For each test seed, calculate the relative difference between its content and that of the standard seed; According to the relative differences, the contents of each substance in the seeds were divided into two groups; Based on the grouping of various indicators and the classification of growth trends, the seeds are divided into several quality levels.

[0013] The beneficial effects of the present invention are as follows: a deep learning model is used to perform image recognition on seeds to be tested, the seed type is determined, and the spectrometer parameters are automatically adjusted, thereby ensuring that the collected spectral data is more accurate and reliable; near-infrared spectral data of seeds are collected, and the internal structure and density information of seeds are obtained through X-ray imaging equipment and ultrasonic detection equipment to form a multimodal data set, thereby improving the richness of data and the predictive ability of the model; a seed vitality prediction model is constructed, and machine learning algorithms such as random forests are used to predict the germination rate of seeds, and the growth trend of seeds is judged according to the prediction results, thereby realizing a comprehensive evaluation of seed quality; seeds are graded according to their growth trends, providing a scientific basis for seed screening and planting management. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0015] Figure 1 This is a flow chart of the method for rapid and non-destructive detection of seed quality based on near-infrared spectroscopy technology in Example 1.

[0016] Figure 2 This is a flow chart for the classification of growth trends in Example 1. DETAILED DESCRIPTION

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0018] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0019] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0020] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for rapid and non-destructive detection of seed quality based on near infrared spectroscopy technology, comprising the following steps: S1, using the deep learning model to perform image recognition on the seeds to be tested, determine the seed type and automatically adjust the spectrometer parameters, including the following steps, A high-resolution digital camera is used to capture images of seed samples in a sample tray; the captured images are grayscaled, denoised, and edge detected; an image segmentation algorithm is used to separate the processed seed images from the background; a convolutional neural network is used to identify the segmented seed images and determine the seed type; and the light source intensity and wavelength range parameters of the spectrometer are automatically adjusted according to the identified seed type.

[0021] Specifically, the specific method of image segmentation is to use the Otsu algorithm to automatically determine the optimal threshold to separate the seed area from the background area. The core idea of ​​the Otsu algorithm is to maximize the inter-class variance, that is, to maximize the variance between the foreground (seed) and the background. The specific steps are as follows: Count the number of pixels at each gray level in the image to form a histogram; Calculate the cumulative probability and between-class variance for each gray level; The expression of the cumulative probability of gray level is: , ;in, Indicates grayscale level and the following cumulative probabilities, i.e., the cumulative probabilities of the foreground (seeds), Indicates grayscale level The number of pixels, Represents the total number of pixels in the image, which is used to normalize the cumulative probability. represents the gray level index, represents the gray level currently considered, Indicates grayscale level The cumulative probability above is the cumulative probability of the background.

[0022] The expression of the between-class variance of gray levels is: ; in, Indicates grayscale level The inter-class variance is used to measure the difference between foreground and background. represents the global average gray value of the image, Indicates grayscale level and the foreground average gray value below; Traverse all gray levels and select the gray level with the largest inter-class variance as the optimal threshold ; According to the optimal threshold, the image is binarized. The specific method is to convert the gray value greater than the optimal threshold The pixels whose gray value is less than or equal to the optimal threshold are set as foreground (seed). The pixels are set as background; The binary expression is: ; in, represents the binarized image, where and are the pixel coordinates, Represents a grayscale image Gray value of the position; Perform morphological processing on the binarized image to remove small noise points and fill holes. The specific method is to use an opening operation (erode first and then expand) to remove small noise points, and use a closing operation (expand first and then erode) to fill holes. The expression of the opening operation is: ; in, represents the image after opening operation, represents the input image, Represents a structural element, usually a small rectangle or circle, Represents the corrosion operation, which is used to remove small noise points or fill holes. Represents the dilation operation, which is used to restore the shape of the image; The expression of the closing operation is: ; in, Represents the image after closing operation; The segmentation results were visually inspected to ensure that the seed region was properly separated from the background region and there was no obvious mis-segmentation. The quality of the segmentation results was evaluated using quantitative metrics such as precision, recall, and F1 score.

[0023] Specifically, the parameter adjustment method is to use a pre-established mapping table between seed types and spectral parameters to find the corresponding light source intensity and wavelength range. The reason for choosing this method is that it can dynamically adjust parameters according to the seed type and improve the quality of spectral data.

[0024] S2, spectral data include moisture content, starch content, fat content and amino acid content of seeds; preprocessing includes moving average, baseline correction, normalization, standard normal variable transformation and detrending processing.

[0025] Specifically, moving average refers to selecting a window length and performing moving average processing on each spectral data point to smooth the spectral data and reduce noise; baseline correction refers to eliminating baseline drift in spectral data by minimizing the original spectral data; normalization refers to using the minimum-maximum normalization method to normalize the spectral data to the same scale to eliminate the influence of dimension; standard normal variable transformation refers to performing a standard normal variable transformation on each spectral data point, which can eliminate the multiple scattering effect in the spectral data and improve the consistency and stability of the spectral data; detrending refers to fitting a linear trend through the least squares method and then subtracting the trend from the spectral data, which can eliminate the linear trend in the spectral data and improve the stability of the spectral data.

[0026] S3, obtaining the internal structure image and density information of the seeds and combining them with the spectral data to form a multimodal data set includes the following steps: Use standard samples to calibrate the X-ray imaging equipment and ultrasonic detection equipment; start the calibrated X-ray imaging equipment to collect the internal structure image of the seeds; start the ultrasonic detection equipment to collect the internal density of the seeds; number the preprocessed spectral data, internal structure image data and internal density data respectively, and match them according to the sample number; create a data table containing the sample number, near-infrared spectral data, X-ray imaging data and ultrasonic detection data; use the sample number as the key to merge the three types of data into the same row to form a multimodal data set and insert it into the data table.

[0027] S4, constructing a seed vitality prediction model, predicting the seed germination rate includes the following steps: Select representative seed samples, use high-throughput sequencing technology to sequence the genome of the seeds and obtain the whole genome sequence; use bioinformatics tools to annotate the sequencing data and identify genes related to germination; extract RNA from seeds at different germination stages; use high-throughput sequencing technology to determine the expression of genes during the germination process and obtain transcriptome data; use bioinformatics tools to process the high-throughput sequencing data and quantitatively analyze the changes in gene expression at different time points; apply statistical methods to identify genes that are significantly upregulated and downregulated during the germination process; select biomarkers that are strongly correlated with germination rate from the differentially expressed genes; fuse the selected biomarkers with near-infrared spectroscopy data, X-ray imaging data, and ultrasonic detection data to form a comprehensive data set, and assign a unique number to each sample; use feature selection algorithms to further screen out the features that have the greatest impact on germination rate and predict the germination rate of seeds, the expression is: ;in, represents the predicted germination rate of seeds, Indicates The comprehensive characteristic value, represents the average value of the comprehensive eigenvector, represents the standard deviation of the comprehensive eigenvector, represents the weight parameter, represents the bias term, Represents the index variable used to identify each comprehensive feature in the comprehensive feature vector.

[0028] Furthermore, high-throughput sequencing technology is used to analyze RNA molecules in cells, which can provide information about gene expression, transcript structure and RNA variation. The specific steps are as follows: Total RNA is extracted from cell or tissue samples, reverse transcribed into cDNA (complementary DNA), fragmented, and sequencing adapters are added; the cDNA library is sequenced using a sequencing platform (such as Illumina, Ion Torrent, etc.); the sequencing data is quality controlled and filtered, and bioinformatics tools are used for alignment and quantitative analysis to identify differentially expressed genes; gene expression patterns are analyzed to identify genes associated with specific biological processes, diseases, or conditions.

[0029] S5, judging the growth trend of the seeds according to the prediction results includes the following steps: According to the predicted germination rate, the growth trend of seeds is divided into three categories: high growth potential, medium growth potential and low growth potential; the growth threshold of germination rate is set as and ,in Indicates the lower limit of high growth potential, Indicates the lower limit of mesopotamia growth potential; when When When When , it is judged as low growth potential.

[0030] S6, grading the seeds according to their growth trends comprises the following steps: The various content data of the test seeds, including water content, protein content, fat content, etc., are read from the database; the various content data of the standard seeds are read from the database as a reference standard; for each test seed, the relative difference between its various contents and the standard seeds is calculated; according to the relative difference, the various contents of the seeds are divided into two groups; based on the grouping of various indicators and the growth trend classification, the seeds are divided into multiple quality levels.

[0031] Furthermore, the relative difference between each content and the standard seed is expressed as: ;in, Indicates The relative difference of the content index is Indicates the test seed Content index, Indicates the standard seed A content indicator.

[0032] The expressions for multiple quality levels of the seed are: ;in, Indicates the quality level of the seed. Indicates the largest relative difference among all indicators. Indicates The weight of each content indicator, Indicates the number of content indicators, represents the value of the growth trend classification, represents the weight of the growth trend classification; Quality Level The value range is 1 to 5, representing different quality levels. Quality Level , indicating that the seed is of the highest quality, all indicators are within the standard range, and the growth trend is high growth potential; quality level , indicating that the seed quality is good, one index exceeds the standard range, and the growth trend is medium growth; quality level , indicating that the seed quality is average, two indicators are beyond the standard range, and the growth trend is low growth potential; quality level , indicating that the seed quality is poor, 3 indicators are beyond the standard range, and the growth trend is low growth potential; quality level , 4 or more indicators are beyond the standard range, and the growth trend is low growth potential.

[0033] This embodiment also provides a computer device, which is suitable for the method of rapid and non-destructive detection of seed quality based on near-infrared spectroscopy technology, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the method of rapid and non-destructive detection of seed quality based on near-infrared spectroscopy technology as proposed in the above embodiment.

[0034] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0035] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing rapid and non-destructive detection of seed quality based on near-infrared spectroscopy technology proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0036] In summary, the present invention: uses a deep learning model to perform image recognition on the seeds to be tested, determines the seed type and automatically adjusts the spectrometer parameters, thereby ensuring that the collected spectral data is more accurate and reliable; collects near-infrared spectral data of seeds, and obtains the internal structure and density information of seeds through X-ray imaging equipment and ultrasonic detection equipment to form a multimodal data set, thereby improving the richness of data and the predictive ability of the model; constructs a seed vigor prediction model, uses machine learning algorithms such as random forests to predict the germination rate of seeds, and judges the growth trend of seeds based on the prediction results, thereby achieving a comprehensive evaluation of seed quality; and grades seeds according to their growth trends, thereby providing a scientific basis for seed screening and planting management.

[0037] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a method for rapid and nondestructive detection of seed quality based on near-infrared spectroscopy technology are provided.

[0038] The experiment selected seeds of three common crops, corn, soybean and wheat, as research objects, aiming to demonstrate the superiority of the present invention by comparing the results of the experimental group with those of the control group. The experiment first prepared necessary instruments and software tools such as a high-resolution digital camera, a near-infrared spectrometer, an X-ray imaging device, an ultrasonic detection device and a random forest model. In order to ensure the accuracy and reliability of the experiment, all equipment was strictly calibrated before use.

[0039] At the beginning of the experiment, a high-resolution digital camera was used to capture images of the seeds to be tested placed in the sample tray. Subsequently, a series of preprocessing operations such as grayscale, denoising and edge detection were performed on the collected original images to improve the quality of the image. Next, an image segmentation algorithm was used to separate the processed seed image from the background in order to more accurately identify the seed type. The segmented image was deeply learned through a convolutional neural network, and the specific type of seed was successfully identified. Based on this, the light source intensity and wavelength range parameters of the near-infrared spectrometer were automatically adjusted to adapt to different types of seeds.

[0040] After adjusting the parameters, the spectral data of the seeds were collected using a near-infrared spectrometer, and a number of preprocessing measures including moving average, baseline correction, and normalization were performed to eliminate the influence of external factors on the data. At the same time, the internal structure images and density information of the seeds were obtained through X-ray imaging equipment and ultrasonic testing equipment, respectively, further enriching the information dimension of the seeds. After preprocessing, these data were accurately matched with the spectral data according to the sample number to form a multimodal data set containing multiple information.

[0041] In the process of building a seed vigor prediction model, the experimental team first calculated the eigenvalues ​​and eigenvectors of the covariance matrix of the spectral data, and selected the first few most important eigenvectors to form the principal component matrix. In addition, convolutional neural networks were used to extract structural features from X-ray images, and support vector machines were used to map the density data detected by ultrasound to a higher-dimensional space. The above three features were fused into a comprehensive feature vector through the feature splicing method. On this basis, a large number of seed samples with known germination rates were collected as training sets and trained using a random forest model. Finally, the comprehensive feature vector was input into the trained model to achieve accurate prediction of seed germination rates.

[0042] According to the predicted germination rate, the growth trend of seeds is divided into three levels: high, medium and low. In addition, the experiment also classifies seeds in detail based on their growth trend and the relative difference of various contents, thus providing scientific guidance for agricultural production.

[0043] The details are shown in Table 1 below: Table 1 Experimental data record table Sample No. Experimental corn Soybean Wheat Corn Control group soybean Wheat Germination rate (%) 89.2 78.5 91.4 75.3 62.4 79.8 Moisture content (%) 13.2 10.5 12.1 15.8 14.2 14.5 Starch content (%) 68.4 57.8 69.2 61.5 52.3 60.1 Fat content (%) 3.8 21.5 2.2 4.1 18.3 2.5 Amino acid content (%) 10.5 8.9 13.4 9.2 7.6 11.2 Internal density (g / cm³) 0.68 0.65 0.71 0.62 0.59 0.66 From the table above, we can see that there are significant differences between the experimental group and the control group in many key indicators such as seed germination rate, moisture content, starch content, fat content, amino acid content and internal density. The germination rates of corn, soybean and wheat in the experimental group were 89.2%, 78.5% and 91.4%, respectively, which were significantly higher than 75.3%, 62.4% and 79.8% in the control group. This result shows that the method of the present invention can more accurately predict the germination ability of seeds, which helps to screen high-quality seeds for planting and improve the efficiency and yield of agricultural production.

[0044] Moisture content: The moisture content of the experimental group is generally lower than that of the control group. For example, the moisture content of corn is 13.2%, while that of the control group is 15.8%. Lower moisture content means that the seeds are better stored and less susceptible to microorganisms such as mold, which prolongs the service life of the seeds.

[0045] Starch content: The starch content of the experimental group is generally higher, such as the starch content of wheat reaching 69.2%, much higher than the 60.1% of the control group. Starch is one of the important sources of energy for plant growth, and a higher starch content is beneficial to promoting seed germination and seedling growth.

[0046] Fat content: Although fat content is not as important as water and starch for seeds, the fat content of the experimental group also showed certain advantages. For example, the fat content of soybeans reached 21.5%, 3.2 percentage points higher than that of the control group. The increase in fat content not only helps to increase the energy density of seeds, but also enhances the cold resistance of seeds, making them more suitable for sowing in cold regions.

[0047] The amino acid content of the experimental group was also significantly higher than that of the control group. For example, the amino acid content of wheat was 13.4%, while that of the control group was only 11.2%. Amino acids are the basic building blocks of protein and are essential for seed development. A higher amino acid content can promote the normal development of seeds and improve the quality of crops.

[0048] Internal density: The internal density of the seeds in the experimental group is generally higher than that in the control group. For example, the internal density of corn is 0.68g / cm³, while that of the control group is 0.62g / cm³. A higher internal density means that the seed structure is more compact, which reduces the chance of invasion by pests and diseases, and is also conducive to the stable growth of seeds in the soil.

[0049] In summary, the method of rapid non-destructive detection of seed quality based on near-infrared spectroscopy technology can not only effectively predict the germination rate of seeds, but also comprehensively evaluate the intrinsic quality of seeds, providing important technical support for agricultural production. Compared with traditional methods, the present invention not only improves the accuracy and efficiency of detection, but also realizes non-destructive detection of seeds, avoiding the waste of resources caused by destructive detection.

[0050] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for rapid and nondestructive detection of seed quality based on near infrared spectroscopy technology, characterized in that: include, Use deep learning models to perform image recognition on seeds to be tested, determine seed type and automatically adjust spectrometer parameters; The spectral data were collected using the adjusted near-infrared spectrometer and preprocessed; Obtain the internal structure and density information of seeds and combine it with spectral data to form a multimodal dataset; Construct a seed vitality prediction model to predict the germination rate of seeds; Determine the growth trend of seeds based on the prediction results; Seeds are graded according to their growth trends.

2. A method for rapid and nondestructive detection of seed quality based on near infrared spectroscopy technology as claimed in claim 1, characterized in that: Using a deep learning model to perform image recognition on the seeds to be tested, determine the seed type and automatically adjust the spectrometer parameters includes the following steps: Use a high-resolution digital camera to capture images of seed samples in the sample tray; Grayscale, denoise and edge detect the collected images; Count the number of pixels at each gray level in the image to form a histogram, and calculate the cumulative probability and inter-class variance of each gray level; Traverse all gray levels and select the gray level with the largest inter-class variance as the optimal threshold; Binarize the image according to the optimal threshold; Perform morphological processing on the binarized image to remove small noise points and fill holes; Use convolutional neural network to identify the segmented seed image and determine the seed type; According to the identified seed type, the light source intensity and wavelength range parameters of the spectrometer are automatically adjusted.

3. A method for rapid and nondestructive detection of seed quality based on near infrared spectroscopy technology as claimed in claim 2, characterized in that: The spectral data include moisture content, starch content, fat content and amino acid content of seeds; the preprocessing includes moving average, baseline correction, normalization, standard normal variable transformation and detrending processing.

4. A method for rapid and nondestructive detection of seed quality based on near infrared spectroscopy technology as claimed in claim 3, characterized in that: Obtaining the internal structure image and density information of seeds and combining them with spectral data to form a multimodal dataset includes the following steps: Calibrate X-ray imaging equipment and ultrasonic testing equipment using standard samples; Start the calibrated X-ray imaging device to collect images of the internal structure of the seeds; Start the ultrasonic testing equipment to collect the internal density of the seeds; The preprocessed spectral data, internal structure image data and internal density data are numbered respectively, and matched according to the sample numbers; Create a data table containing sample ID, near infrared spectroscopy data, X-ray imaging data, and ultrasonic detection data; Using the sample number as a key, the three data were merged into the same row to form a multimodal dataset and inserted into the data table.

5. A method for rapid and nondestructive detection of seed quality based on near infrared spectroscopy technology as claimed in claim 4, characterized in that: Constructing a seed vigor prediction model to predict the germination rate of seeds includes the following steps: Select representative seed samples, use high-throughput sequencing technology to sequence the seed genome, and obtain the whole genome sequence; Bioinformatics tools were used to annotate the sequencing data and identify genes associated with germination; RNA was extracted from seeds at different germination stages; High-throughput sequencing technology was used to measure the expression of genes during germination and obtain transcriptome data; Use bioinformatics tools to process high-throughput sequencing data and quantitatively analyze changes in gene expression at different time points; Apply statistical methods to identify genes that are significantly up-regulated and down-regulated during germination. Biomarkers strongly correlated with germination rate were selected from the differentially expressed genes; The selected biomarkers were fused with near-infrared spectroscopy data, X-ray imaging data, and ultrasound detection data to form a comprehensive data set, and a unique number was assigned to each sample; The feature selection algorithm is used to further screen out the features that have the greatest impact on the germination rate and predict the germination rate of seeds. The expression is: ;in, represents the predicted germination rate of seeds, Indicates The comprehensive characteristic value, represents the average value of the comprehensive eigenvector, represents the standard deviation of the comprehensive eigenvector, represents the weight parameter, represents the bias term, Represents the index variable used to identify each comprehensive feature in the comprehensive feature vector.

6. A method for rapid and nondestructive detection of seed quality based on near infrared spectroscopy technology as claimed in claim 5, characterized in that: According to the prediction results, judging the growth trend of seeds includes the following steps: According to the predicted germination rate, the growth trend of seeds is divided into three categories: high growth potential, medium growth potential and low growth potential; The growth threshold of germination rate is set to and ,in Indicates the lower limit of high growth potential, Indicates the lower limit of mesopotamia growth potential; when When When When , it is judged as low growth potential.

7. A method for rapid and nondestructive detection of seed quality based on near infrared spectroscopy technology as claimed in claim 6, characterized in that: Grading seeds according to their growth trends includes the following steps: Read various content data of the test seeds and various content data of the standard seeds from the database as reference standards; For each test seed, calculate the relative difference between its content and that of the standard seed; According to the relative differences, the contents of each substance in the seeds were divided into two groups; Based on the grouping of various indicators and the classification of growth trends, the seeds are divided into several quality levels.

Images