Method and system for rapidly identifying photo-thermo-sensitive sterile line of forage grass sorghum

Abstract

The invention discloses a method and a system for rapidly identifying a photo-thermo-sensitive sterile line of forage grass sorghum, and relates to the technical field of agricultural remote sensing. The method comprises the following specific steps: taking a suspected photo-thermo-sensitive sterile line as a material to be detected; acquiring hyperspectral data of the to-be-detected material after heading and before pollen formation by using a hyperspectral imaging technology; extracting identification characteristics related to the photo-thermo-sensitive sterile line of the forage grass sorghum from the hyperspectral data; and establishing an identification model based on the identification characteristics, and inputting a to-be-detected sample into the identification model for rapid identification of the photo-thermo-sensitive sterile line. The method is based on a hyperspectral imaging technology and is combined with a machine learning algorithm to realize rapid, lossless, visual and accurate identification of the photo-thermo-sensitive sterile line of the forage grass sorghum, and the method can greatly improve the identification efficiency and accuracy.

Description

Technical Field

[0001] The invention relates to the technical field of agricultural remote sensing, and in particular to a method and system for quickly identifying a forage sorghum photothermosensitive sterile line. Background Art

[0002] Forage sorghum refers to sorghum whose above-ground parts are harvested as a whole plant for forage, mainly including sweet sorghum, sudan grass and sorghum grass. As a high-quality forage, forage sorghum has the characteristics of high grass yield, good feeding quality and outstanding stress resistance. In recent years, it has gradually been favored by breeders and farmers in production. Through research, it is found that the use of the "two-line method" to utilize hybrid advantages simplifies the breeding process and overcomes the problem of narrow selection of restorer line resources in "three-line" hybridization. It is a convenient way to utilize the hybrid advantages of forage sorghum. The sterile line in the "two-line" mainly refers to the photothermosensitive sterile line, which has the characteristics of one line for two purposes. Its fertility changes are mainly affected by ecological conditions such as temperature or photoperiod. Under specific ecological conditions, it can be used as a sterile line, and can also be self-bred to breed sterile seeds. The commonly used methods for identifying photo-thermo-sensitive sterile lines are: potted planting method, different sowing periods, off-site planting method, backcrossing, and hybridization method; all of the above methods require the naked eye or a microscope to assist in observing whether there is pollen in the pollen sac after the pollen sac is formed, or whether the pollen has fertilization ability. They are not fast and accurate enough in identifying photo-thermo-sensitive sterile lines of forage sorghum. For those skilled in the art, how to quickly identify photo-thermo-sensitive sterile lines of forage sorghum efficiently and accurately is an urgent problem to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for quickly identifying photothermosensitive sterile lines of forage sorghum, so as to solve the problems raised in the background technology.

[0004] To achieve the above object, the present invention provides the following scheme: a method for rapid identification of forage sorghum photothermosensitive sterile lines, the specific steps of which include the following:

[0005] The suspected photo-thermo-sensitive sterile lines were used as materials to be tested;

[0006] Using hyperspectral imaging technology to obtain hyperspectral data of the material to be inspected from heading to pollen formation;

[0007] extracting identification features associated with forage sorghum photothermosensitive male sterile lines from the hyperspectral data;

[0008] A recognition model is established based on the identification characteristics, and the sample to be tested is input into the recognition model for rapid identification of photothermosensitive sterile lines.

[0009] Preferably, the method further includes performing a preprocessing operation on the hyperspectral data, wherein the preprocessing operation includes performing a de-mosaicing process on the hyperspectral data, performing a spectral image resolution enhancement process on the hyperspectral data, and performing a filtering and denoising process on the hyperspectral data.

[0010] Preferably, forage sorghum samples are collected under different conditions, and suspected photothermosensitive sterile strains found under short-day and low-temperature conditions are used as materials to be tested.

[0011] Preferably, a hyperspectral sorter is used to obtain a hyperspectral image of the material to be inspected from heading to pollen formation, the hyperspectral image is segmented, and after masking, the average value of all spectra in the masked area is taken to obtain the hyperspectral data.

[0012] Preferably, the specific steps of constructing the recognition model are:

[0013] Establishing a convolutional neural network model, selecting part of the hyperspectral data for labeling through the identification features, and storing the data as samples in a sample database;

[0014] The sample database is divided into a training set and a test set, and the convolutional neural network model is trained with the training set so that it can identify different types of forage sorghum plants, including photothermosensitive sterile lines and normal lines;

[0015] The trained convolutional neural network model is verified by the test set, and finally the recognition model is obtained.

[0016] Preferably, the method also includes using the regeneration ability of forage sorghum rhizomes to verify the stable genetic characteristics of its light and temperature sensitivity.

[0017] Preferably, the identification feature is extracted by convolving the optimal band feature image with an adaptive random block convolution kernel through a hierarchical network, and the specific steps are as follows:

[0018] Step 1: First, set the number of layers of the hierarchical network;

[0019] Step 2, extracting the characteristics of the photothermosensitive sterile line of forage sorghum in the first layer according to the adaptive random block convolution kernel and the optimal band feature image convolution;

[0020] Step 3: principal component analysis is used to reduce the dimension of the first layer of forage sorghum photothermosensitive sterile line features, a random projection method is used to obtain random blocks as convolution kernels from the reduced-dimensional data, and a pixel adaptive method is used to modify the convolution kernel to obtain an adaptive random block convolution kernel of the second layer;

[0021] Step 4, performing convolution extraction based on the adaptive random block convolution kernel of the second layer and the features of the forage sorghum photo-thermo-sensitive male sterile line in the first layer to obtain the features of the forage sorghum photo-thermo-sensitive male sterile line in the second layer;

[0022] Step 5, repeat step 4 until the characteristics of the photothermosensitive sterile line of forage sorghum in the nth layer are extracted as the identification characteristics.

[0023] On the other hand, a rapid identification system for forage sorghum photothermosensitive sterile lines is provided, comprising a data acquisition module, a hyperspectral imaging module, a feature extraction module, and an identification module; wherein:

[0024] The data acquisition module is used to use the suspected photo-thermo-sensitive sterile strains as materials to be tested;

[0025] The hyperspectral imaging module is used to obtain hyperspectral data of the material to be inspected from heading to pollen formation using hyperspectral imaging technology;

[0026] The feature extraction module is used to extract identification features related to the photothermosensitive male sterile line of forage sorghum from the hyperspectral data;

[0027] The identification module is used to establish an identification model based on the identification features, and input the sample to be tested into the identification model for rapid identification of the photothermosensitive sterile line.

[0028] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: based on hyperspectral imaging technology, combined with machine learning algorithms, rapid, non-destructive, intuitive and accurate identification of forage sorghum photothermosensitive sterile lines can be achieved, which can greatly improve the efficiency and accuracy of identification and reduce the complexity and time cost of manual operation. The advantage of hyperspectral imaging technology is that it can provide rich spectral information, which helps to improve the accuracy and efficiency of identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0030] Figure 1 is a flow chart of the method of the present invention;

[0031] Figure 2 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] The object of the present invention is to provide a method for rapid identification of photothermosensitive male sterile lines of forage sorghum. Figure 1 As shown, the specific steps include the following:

[0034] S1. Use the suspected photo-thermo-sensitive sterile strains as the materials to be tested;

[0035] S2, using hyperspectral imaging technology to obtain hyperspectral data of the material to be tested from heading to pollen formation;

[0036] S3, extracting identification features related to photothermosensitive male sterile lines of forage sorghum from hyperspectral data;

[0037] S4. Establish a recognition model based on the identification features, and input the sample to be tested into the recognition model for rapid identification of the photothermosensitive sterile line.

[0038] Hyperspectral imaging technology uses an imaging spectrometer to obtain continuous and subdivided spectral bands in the ultraviolet, visible light, near-infrared and mid-infrared regions of the electromagnetic spectrum, forming a three-dimensional data block, including two-dimensional plane pixel information and one-dimensional wavelength information. This technology can fully reflect the differences in physical structure and chemical composition inside the sample.

[0039] Furthermore, hyperspectral imaging was performed after the sensitive period of fertility conversion of forage sorghum and before pollen formation. In the present embodiment, hyperspectral data was collected after heading and before pollen formation. The hyperspectral imaging experiment was run in a darkroom, and spectral information of 462 bands (392.38-1011.01 nm) was collected. For each sample, the image region of interest (ROI) was manually selected using ENVI5.3 software, and the spectral average of all pixels in the ROI was its spectral reflectance value. Extracting the spectrum using the entire sample as the region of interest can more completely and comprehensively represent the internal component information of sorghum, which is conducive to subsequent classification and identification.

[0040] In certain specific wavelength ranges, the spectral reflectance of photothermosensitive sterile forage sorghum is different from that of fertile forage sorghum. For example, in the near-infrared band, photothermosensitive sterile forage sorghum will have some unique absorption peaks or absorption valleys. Different biochemical components will produce specific spectral responses at specific wavelengths. Due to changes in its physiological characteristics, the content and structure of certain biochemical components in photothermosensitive sterile sorghum will change, thus showing different spectral characteristics from fertile sorghum in hyperspectral imaging.

[0041] It should be noted that the use of hyperspectral imaging technology focuses on measuring the following agronomic traits of forage sorghum from heading to pollen formation: leaf morphology and ear characteristics. Hyperspectral imaging can capture the texture and tissue structure information of the leaves. When the leaves of the photothermo-sensitive sterile sorghum are in a sterile state, the cell structure will change. These changes will be reflected in hyperspectral imaging as changes in the clarity and roughness of the leaf texture, as well as changes in the scattering and absorption characteristics of the internal tissue structure of the leaves to light.

[0042] The pigments in leaves have an important influence on the absorption and reflection of light. Hyperspectral imaging can detect the distribution and content changes of pigments in leaves. Under different light and temperature conditions, the content and proportion of chlorophyll, carotenoids and other pigments in the leaves of photothermosensitive male sterile sorghum will change, resulting in changes in leaf color.

[0043] The ear of sorghum is an important reproductive organ. Before the fertility of photothermo-sensitive sterile sorghum can be distinguished by naked eyes, the shape, size, color and other characteristics of the ear in the sterile state are different from those of the fertile sorghum ear, but these morphological differences are difficult to observe with the naked eye. Hyperspectral imaging can analyze the spectral characteristics of the ear, thereby assisting in judging the fertility status of sorghum.

[0044] Furthermore, the acquisition of hyperspectral signals will be affected by factors such as the stability of the hyperspectral experimental system and the experimental environment. Therefore, it is necessary to preprocess the acquired hyperspectral data to improve the correlation between the spectral information and the internal chemical composition. The preprocessing operation includes de-mosaicing of the hyperspectral data, enhancing the spectral image resolution of the hyperspectral data, and filtering and denoising the hyperspectral data.

[0045] Furthermore, forage sorghum samples were collected under different conditions, and suspected photothermosensitive sterile lines were found under short-day and low-temperature conditions as materials to be tested. A hyperspectral sorter was used to obtain hyperspectral images of the materials to be tested from heading to pollen formation, and the hyperspectral images were segmented. After masking, the average value of all spectra in the masked area was taken to obtain hyperspectral data.

[0046] Furthermore, the hierarchical network is used to convolve the optimal band feature image using an adaptive random block convolution kernel to extract the identification features. The specific steps are as follows:

[0047] S31, first set the number of layers of the hierarchical network;

[0048] S32, extracting the characteristics of the photothermosensitive sterile line of forage sorghum in the first layer according to the adaptive random block convolution kernel and the optimal band feature image convolution;

[0049] S33, using principal component analysis to perform data dimensionality reduction on the characteristics of the first layer of forage sorghum photothermosensitive sterile lines, using a random projection method to obtain random blocks as convolution kernels from the reduced-dimensional data, and using a pixel adaptive method to modify the convolution kernel to obtain an adaptive random block convolution kernel of the second layer;

[0050] S34, performing convolution extraction based on the adaptive random block convolution kernel of the second layer and the characteristics of the forage sorghum photo-thermo-sensitive sterile line in the first layer to obtain the characteristics of the forage sorghum photo-thermo-sensitive sterile line in the second layer;

[0051] S35, repeat S34 until the characteristics of the forage sorghum photo-thermo-sensitive sterile line in the nth layer are extracted as the identification characteristics.

[0052] Furthermore, the specific steps of building the recognition model are:

[0053] S41, establishing a convolutional neural network model, selecting part of the hyperspectral data for labeling through identification features, and storing it in a sample database as a sample;

[0054] S42, dividing the sample database into a training set and a test set, and using the training set to train the convolutional neural network model so that it can identify different types of forage sorghum plants, including photothermosensitive sterile lines and normal lines;

[0055] S43. Verify the trained convolutional neural network model through the test set and finally obtain the recognition model.

[0056] Furthermore, it also includes using the regeneration ability of forage sorghum rhizomes to verify the stable genetic characteristics of its light and temperature sensitivity. The specific steps are: the light and temperature sensitive sterile line materials of forage sorghum identified by hyperspectral imaging technology will be sown in Sanya, Hainan in early November 2023. The sterile line will show sterility characteristics in February and March of the following year. Taking advantage of the fact that the rhizomes of forage sorghum can be regenerated after mowing, the individual plants are mowed, and the fertility recovery of the regenerated stems of the same individual plant under different light and temperature conditions at the same location is observed. It is found that the regenerated stems of the individual plant can bear fruit normally and the fertility is restored in Sanya, Hainan in June. In order to further verify the fertility of the individual plant under light and temperature conditions that are basically consistent with its regenerated stems, the rhizomes of the individual plant are transplanted to Hengshui (the light and temperature conditions in Hengshui after transplantation are basically consistent with the light and temperature conditions during the growth of the regenerated stems in Sanya, Hainan), and the fertility recovery of the regenerated stems is further observed. In October, the fertility of the regenerated stems was also found to have been restored in Hengshui. Under the condition of the same genotype of the same single plant, the root and stem regeneration ability of forage sorghum was accurately verified three times, which confirmed the feasibility of hyperspectral imaging technology in identifying forage sorghum photothermosensitive sterile lines, and the identification was accurate. The constructed identification model can be used for a long time, improving the efficiency of identification and saving costs. The use of hyperspectral imaging technology to identify forage sorghum photothermosensitive sterile lines is earlier than the naked eye observation of its sterility, which can provide sufficient time for the use of the sterile lines and improve breeding efficiency.

[0057] On the other hand, a rapid identification system for forage sorghum photothermosensitive sterile lines is provided. Figure 2 As shown, it includes a data acquisition module, a hyperspectral imaging module, a feature extraction module, and an identification module; wherein,

[0058] A data acquisition module, used to use the suspected photothermosensitive sterile strains as materials to be tested;

[0059] A hyperspectral imaging module, used for acquiring hyperspectral data of the material to be inspected by using hyperspectral imaging technology;

[0060] A feature extraction module is used to extract the identification features related to the photothermosensitive male sterile lines of forage sorghum from the hyperspectral data;

[0061] The identification module is used to establish an identification model based on the identification features, and input the sample to be tested into the identification model for rapid identification of the photothermosensitive sterile line.

[0062] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for rapid identification of photothermosensitive male sterile lines of forage sorghum, characterized in that: The specific steps include the following: The suspected photo-thermo-sensitive sterile lines were used as materials to be tested; Using hyperspectral imaging technology to obtain hyperspectral data of the material to be inspected from heading to pollen formation; extracting identification features associated with forage sorghum photothermosensitive male sterile lines from the hyperspectral data; A recognition model is established based on the identification characteristics, and the sample to be tested is input into the recognition model for rapid identification of photothermosensitive sterile lines.

2. The method for rapid identification of forage sorghum photo-thermo-sensitive male sterile lines according to claim 1, characterized in that: The method also includes performing a preprocessing operation on the hyperspectral data, wherein the preprocessing operation includes performing a demosaicing process on the hyperspectral data, performing a spectral image resolution enhancement process on the hyperspectral data, and performing a filtering and denoising process on the hyperspectral data.

3. The method for rapid identification of forage sorghum photo-thermo-sensitive male sterile lines according to claim 1, characterized in that: Forage sorghum samples were collected under different conditions, and suspected photothermosensitive sterile lines were found under short-day and low temperature conditions as materials for inspection.

4. The method for rapid identification of forage sorghum photo-thermo-sensitive male sterile lines according to claim 1, characterized in that: A hyperspectral sorter is used to obtain a hyperspectral image of the material to be inspected from heading to pollen formation, the hyperspectral image is segmented, and after masking, the average value of all spectra in the masked area is taken to obtain the hyperspectral data.

5. The method for rapid identification of forage sorghum photo-thermo-sensitive male sterile lines according to claim 1, characterized in that: The specific steps of constructing the recognition model are: Establishing a convolutional neural network model, selecting part of the hyperspectral data for labeling through the identification features, and storing the data as samples in a sample database; The sample database is divided into a training set and a test set, and the convolutional neural network model is trained with the training set so that it can identify different types of forage sorghum plants, including photothermosensitive sterile lines and normal lines; The trained convolutional neural network model is verified by the test set, and finally the recognition model is obtained.

6. The method for rapid identification of forage sorghum photo-thermo-sensitive male sterile lines according to claim 1, characterized in that: It also includes the verification of the stable genetic characteristics of the light and temperature sensitive characteristics of forage sorghum using its regeneration ability of rhizomes.

7. The method for rapid identification of forage sorghum photo-thermo-sensitive male sterile lines according to claim 1, characterized in that: The identification features are extracted by convolving the optimal band feature image with an adaptive random block convolution kernel through a hierarchical network. The specific steps are as follows: Step 1: First, set the number of layers of the hierarchical network; Step 2, extracting the characteristics of the photothermosensitive sterile line of forage sorghum in the first layer according to the adaptive random block convolution kernel and the optimal band feature image convolution; Step 3: principal component analysis is used to reduce the dimension of the first layer of forage sorghum photothermosensitive sterile line features, a random projection method is used to obtain random blocks as convolution kernels from the reduced-dimensional data, and a pixel adaptive method is used to modify the convolution kernel to obtain an adaptive random block convolution kernel of the second layer; Step 4, performing convolution extraction based on the adaptive random block convolution kernel of the second layer and the features of the forage sorghum photo-thermo-sensitive male sterile line in the first layer to obtain the features of the forage sorghum photo-thermo-sensitive male sterile line in the second layer; Step 5, repeat step 4 until the characteristics of the photothermosensitive sterile line of forage sorghum in the nth layer are extracted as the identification characteristics.

8. A rapid identification system for forage sorghum photothermosensitive sterile lines, characterized in that: It includes data acquisition module, hyperspectral imaging module, feature extraction module and identification module; among them, The data acquisition module is used to use the suspected photo-thermo-sensitive sterile strains as materials to be tested; The hyperspectral imaging module is used to obtain hyperspectral data of the material to be inspected from heading to pollen formation using hyperspectral imaging technology; The feature extraction module is used to extract identification features related to the photothermosensitive male sterile line of forage sorghum from the hyperspectral data; The identification module is used to establish an identification model based on the identification features, and input the sample to be tested into the identification model for rapid identification of the photothermosensitive sterile line.

Images