A method and equipment for obtaining the comprehensive lodging resistance coefficient of soybeans

Abstract

This invention relates to the fields of intelligent crop phenotypic detection, soybean molecular breeding, and crop stress resistance evaluation, specifically to a method and device for obtaining the comprehensive lodging resistance coefficient of soybeans. Utilizing hyperspectral-3D point cloud deep fusion organ segmentation technology in the same coordinate system, morphological and physiological dual-dimensional organ-level parameters are extracted simultaneously to construct a three-layer fused lodging resistance evaluation index, enabling rapid and non-destructive pre-screening of lodging resistance genotypes in breeding populations. The method involves multi-source image homology registration and organ-level data fusion acquisition, intelligent organ-level point cloud segmentation and simultaneous extraction of dual-dimensional parameters, modeling and calculating the lodging resistance grading index, and pre-screening of breeding population genotypes.

Description

Technical Field

[0001] This invention relates to the fields of intelligent crop phenotypic detection, soybean molecular breeding, and crop stress resistance trait evaluation, specifically to a method and equipment for obtaining the comprehensive lodging resistance coefficient of soybeans. Background Technology

[0002] Soybean lodging is a core stressor restricting high and stable soybean yields, mechanized harvesting, and grain quality. Lodging can directly reduce soybean yield by 15%-40% and significantly increase breeding costs. Currently, there are three major technical shortcomings in the screening of soybean lodging-resistant genotypes and the evaluation of lodging resistance: 1. Traditional evaluation methods are outdated: The mainstream method is to use manual lodging assessment in the field at maturity and mechanical damage detection of stem puncture / bending. The detection efficiency is low, the subjectivity is strong, and the damage detection cannot preserve the live breeding plants, so it cannot be linked to subsequent gene sequencing work. 2. Limited Dimensions of Detection Indicators: Existing optical detection technologies rely solely on appearance indicators such as plant height and stem diameter, neglecting the lodging resistance intrinsic characteristics of physiological components such as lignin and cellulose in the stem. Furthermore, they do not take into account environmental lodging stress factors such as field wind damage and soil moisture, resulting in significant deviations in lodging resistance evaluation results. 3. Insufficient accuracy of phenotypic fusion: Existing hyperspectral and 3D point cloud technologies are used independently without a homologous coordinate coupling mechanism. They can only achieve overall plant detection and cannot separate refined parameters of the main stem and branch organs. Organ-level lodging resistance correlation data are missing, making it impossible to accurately distinguish genotype lodging resistance differences. The fit of large-scale breeding population screening is extremely poor.

[0003] Based on this, the present invention proposes an integrated method and dedicated equipment, filling a technological gap in the industry. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing soybean lodging resistance detection methods, such as single indicators, damaged detection, low efficiency, inability to quantify at the organ level, and difficulty in high-throughput screening for breeding. This invention provides a method and device for obtaining the comprehensive lodging resistance coefficient of soybeans. It relies on the deep fusion organ segmentation technology of hyperspectral and three-dimensional point cloud in the same coordinate system to simultaneously extract organ-level parameters of morphology and physiology, construct a three-layer fused lodging resistance evaluation index, and realize rapid and non-destructive pre-screening of lodging resistance genotypes in breeding populations.

[0005] The objective of this invention is achieved through the following technical solution: I. A method for obtaining the comprehensive lodging resistance coefficient of soybeans It includes the following four core steps: Step 1: Multi-source image registration and deep fusion of dual technologies The external parameters of the hyperspectral imaging module and the 3D laser point cloud module were uniformly calibrated, and a unique world coordinate system for the plant was established to achieve temporal synchronization and pixel homogeneous binding of spectral information and 3D morphology information. The hyperspectral images and 3D original point clouds of the soybean plants to be tested were collected in situ in the field to eliminate the spatial misalignment error of the dual-source data and provide a fused dataset for organ-level segmentation.

[0006] Step 2: Spectral-spatial dual-constraint organ-level point cloud segmentation, with simultaneous extraction of two parameters. The fused dataset underwent denoising and background stripping preprocessing. Leveraging the dual constraints of spectral vegetation features and 3D spatial morphology, a deep learning semantic segmentation model was used to split the point cloud clusters of individual soybean organs—main stem, branches, petioles, and pods—achieving organ-level fine segmentation. Based on the segmented organ-specific point clouds, two types of core parameters were extracted simultaneously and losslessly. 1. Organ-level morphological parameters: These determine the plant's resistance to overturning and bending, including the thickness of the main stem base, stem wall thickness, internode length, plant center of gravity height, branching angle, and three-dimensional volume of the aboveground parts. 2. Organ-level physiological component parameters: These determine the structural strength of the stem material. The stem lignin, cellulose, water content, soluble sugar, and vegetation stress resistance spectral index are obtained by inversion using hyperspectral bands.

[0007] Step 3: Construct a three-layer coupling index model and calculate the comprehensive lodging resistance coefficient. Abandoning the single-index evaluation model, a three-layer independent evaluation factor system is constructed, and a standardized comprehensive lodging resistance coefficient is obtained through weighted fusion: 1. Morphological composite value A: Based on the analytic hierarchy process (AHP) and combined with the weighting assigned by breeding experts, all organ morphological parameters are normalized and integrated to quantify the lodging resistance of the plant's external structure. 2. Physiological composite value B: By correlating physiological components with measured data of stem mechanics, a spectral inversion physiological value-stem bending strength fitting model is established to obtain the plant's endogenous physiological lodging resistance. 3. Environmental stress index C: Real-time data collection of field wind speed, soil waterlogging, planting density, and light stress to quantify the intensity of external environmental induction stress on soybean lodging; Finally, the standardized comprehensive lodging resistance coefficient is calculated using the coupling formula X=αA+βB-γC. The larger the coefficient value, the stronger the lodging resistance of a single soybean plant.

[0008] Step 4: Threshold grading and high-throughput pre-screening of breeding genotypes By combining regional soybean breeding big data to determine grading thresholds, individual plants in the population are divided into three genotypes: highly resistant, moderately resistant, and susceptible to lodging. Germplasm numbers, growth periods, lodging resistance coefficients, and grade data are automatically archived, and susceptible lodging materials are eliminated in batches, completing rapid pre-screening of breeding populations and reducing the workload of subsequent manual identification in the field.

[0009] II. A special device for obtaining the comprehensive lodging resistance coefficient of soybeans The equipment is a tracked intelligent phenotypic data acquisition device for large fields, with five main modules working in tandem: 1. Tracked mobile frame: Suitable for walking between rows in soybean ridge and flat planting breeding plots, with controllable walking accuracy, obstacle avoidance and passage, suitable for high-density breeding population inspection; 2. Coordinated Acquisition Module: Integrates a hyperspectral camera and a 3D lidar, sharing a calibration coordinate system, and synchronously acquires spectral and point cloud data, with an acquisition speed adapted to high-throughput field operations; 3. Field environment sensing module: integrates wind speed sensor, soil moisture sensor, and plant spacing density sensor to upload environmental stress parameters in real time; 4. Industrial Control Intelligent Processing Terminal: Built-in fusion segmentation algorithm, three-layer index modeling algorithm, and breeding grading algorithm, it can complete data processing, coefficient calculation, and genotype grading locally without cloud transmission; 5. Data storage and transmission module: Locally stores breeding germplasm data, can connect to breeding big data platforms, export screening reports, and link with germplasm gene sequencing equipment.

[0010] The beneficial effects of the mobile construction waste treatment equipment of the present invention are as follows: 1. The pioneering hyperspectral-3D point cloud co-coordinate system organ-level fusion segmentation technology breaks through the limitations of traditional whole plant detection, accurately segments soybean functional organs, and obtains morphological and physiological parameters that are consistent with the core mechanism of stem lodging, thus greatly improving the evaluation accuracy; 2. A pioneering three-layer coupled lodging resistance index model combining morphology, physiology, and environment is developed, taking into account the plant's endogenous resistance, external resistance, and external lodging stress. This model differs from single-morphological evaluation, and the lodging resistance assessment results closely match the actual lodging performance in the field. 3. The entire process involves in-situ, non-destructive testing, which does not damage the living soybean plants used in breeding. After testing, the plants can continue to undergo phenotypic observation, gene sequencing, and omics experiments, thus adapting to the entire breeding process. 4. High-throughput and efficient screening, with a single plant detection time of ≤2s, can complete the detection of more than 2,000 breeding germplasm samples per day, which is more than 8 times more efficient than manual identification, and is suitable for pre-screening of large-scale breeding populations; 5. The equipment is integrated into a single unit, and the ecological zone parameters can be iteratively calibrated. It is suitable for breeding in multiple ecological zones, including the Huang-Huai-Hai Plain, Northeast China spring soybean, and Southern China summer soybean, making it highly versatile. Attached Figure Description

[0011] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0012] Figure 1This is a flowchart of the steps for obtaining the comprehensive lodging resistance coefficient of soybeans according to the present invention; Figure 2 This is a schematic diagram of the overall assembly structure of the special acquisition device of the present invention; In the diagram: 1-Mobile data acquisition frame; 2-Same-source data acquisition module; 3-Field environment sensing module; 4-Processing terminal; 5-Data storage and transmission module. Detailed Implementation

[0013] Example 1: Pre-screening of lodging-resistant genotypes in Northeast spring soybean breeding populations 1. Equipment calibration: Complete the extrinsic parameter calibration of the hyperspectral camera (400-1000nm) and the 3D lidar, and calibrate the coordinate error to within 0.2mm; 2. Field data collection: During the soybean grain-filling stage (critical growth period for lodging), the tracked machine moves between rows to simultaneously collect hyperspectral and three-dimensional point cloud data of 1200 breeding lines, as well as environmental data such as field wind speed and soil moisture content. 3. Organ segmentation: The system automatically removes the point cloud of soil weeds, segments the point cloud of main stem and branch organs, extracts morphological parameters such as stem diameter and center of gravity height, and inverts physiological parameters such as lignin and water content; 4. Index Calculation: Substitute the calibration coefficients of Northeast soybean α=0.42, β=0.45, and γ=0.13 to calculate the comprehensive lodging resistance coefficient X of a single plant; Example 2: Lodging resistance identification of summer soybean germplasm resources in the Huang-Huai-Hai Plain The regional weighting coefficients α=0.38, β=0.48, and γ=0.14 were iteratively optimized. For testing of 800 local soybean germplasm accessions, the lodging resistance grade and the artificial lodging rating at maturity showed a 93.6% agreement, meeting the germplasm entry identification standards.

[0014] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention are also within the protection scope of the present invention.

Claims

1. A method for obtaining the comprehensive lodging resistance coefficient of soybeans, characterized in that, Includes the following steps: S1. Multi-source image homology registration and organ-level data fusion acquisition: The hyperspectral imaging module and the three-dimensional laser point cloud acquisition module are linked to the acquisition platform to calibrate the spatial parameters of the two modules and construct a unified coordinate system for soybean plants in the field; the hyperspectral image dataset and the three-dimensional original point cloud dataset of the soybean plants to be tested are acquired simultaneously to complete the homology mapping between hyperspectral pixels and three-dimensional point cloud pixels. S2. Intelligent segmentation of organ-level point clouds and simultaneous extraction of dual-dimensional parameters: Pixel clustering segmentation is performed on the homologous fusion dataset to divide the soybean main stem, branches, petioles and pods into independent point cloud clusters, realizing fine segmentation of plant organs at the organ level; Based on the segmented organ point clouds, organ-level morphological parameters and organ-level physiological component parameters are extracted simultaneously. S3. Modeling and calculating the lodging resistance grading index: Extract parameters to construct three-level evaluation factors for soybean morphological comprehensive value, physiological comprehensive value and field environmental stress index, and solve the comprehensive lodging resistance coefficient of the soybean plant under test; S4. Genotyping and pre-screening of breeding populations: Set a threshold for lodging resistance coefficient grading, grade all soybean plants in the breeding population in batches, and quickly screen germplasm materials with high lodging resistance genotypes, medium lodging resistance genotypes, and lodging susceptibility genotypes to complete the pre-screening of breeding populations.

2. The method for obtaining the comprehensive lodging resistance coefficient of soybeans according to claim 1, characterized in that, The dual-module homogeneous registration method in step S1 is as follows: the extrinsic parameters of the hyperspectral camera and the three-dimensional lidar are calibrated using a checkerboard calibration board, and the spatial coordinate origin, scale factor, and rotation matrix are unified to achieve the same coordinate binding and time-series synchronous acquisition of hyperspectral spectral information and three-dimensional spatial topography information.

3. The method for obtaining the comprehensive lodging resistance coefficient of soybeans according to claim 1, characterized in that, The specific steps of the organ-level point cloud segmentation process in step S2 are as follows: S21. Perform noise reduction, ground point stripping, and plant outlier removal preprocessing on the original 3D point cloud to obtain a pure 3D point cloud of plants. S22. By integrating the threshold values ​​of hyperspectral vegetation feature bands, a spectral-spatial dual-constraint segmentation model is constructed to eliminate background interference from soil and weeds. S23. Based on the trained crop organ semantic segmentation model, classify and segment point by point, and independently output four subdivided point cloud clusters: soybean main stem organ point cloud, branch organ point cloud, petiole organ point cloud, and pod organ point cloud.

4. The method for obtaining the comprehensive lodging resistance coefficient of soybeans according to claim 1, characterized in that, Step S2 extracts specific parameter divisions: Organ-level morphological parameters: main stem thickness, stem wall thickness, plant height, center of gravity height, branching angle, stem internode length, aboveground volume, and organ spatial bending morphology parameters; Organ-level physiological component parameters: stem cellulose content, lignin content, soluble sugar content, water content, plant chlorophyll index, and stress resistance spectrum index.

5. The method for obtaining the comprehensive lodging resistance coefficient of soybeans according to claim 1, characterized in that, Step S3: Three-layer index construction method: S31 Morphological Comprehensive Value: Based on the analytic hierarchy process, the morphological parameters of each organ are assigned weights and weighted normalization is used to calculate the lodging resistance morphological comprehensive value of the plant, which characterizes the plant's spatial structure resistance to bending and overturning. S32 Physiological Comprehensive Value: Based on the content of physiological components obtained by hyperspectral inversion of organs, the physiological comprehensive value of stem structural strength is calculated to characterize the intrinsic ability of plant stem material to resist lodging. S33 Environmental Stress Index: Integrates field wind speed, soil moisture content, planting density, and light stress environmental factors to quantify the intensity of external lodging stress; S34 Comprehensive Lodging Resistance Coefficient Calculation Formula: X = αA + βB - γC In the formula: X is the comprehensive lodging resistance coefficient of soybean; A is the comprehensive morphological value; B is the comprehensive physiological value; C is the environmental stress index; α, β, and γ are the weighting coefficients for breeding scenario calibration.

6. The method for obtaining the comprehensive lodging resistance coefficient of soybeans according to claim 1, characterized in that, Step S4 Breeding Screening and Grading Criteria: I ≥ 0.75 is judged as a highly lodging-resistant genotype; 0.45 ≤ I < 0.75 is judged as a moderately lodging-resistant genotype. Materials with an I < 0.45 were identified as susceptible to lodging. The breeding material numbers, lodging resistance levels, and coefficient data were exported in batches to complete the pre-screening.

7. A device for obtaining the comprehensive lodging resistance coefficient of soybeans, used to implement the acquisition method described in any one of claims 1-6, characterized in that, It includes a mobile data acquisition frame, a homogeneous data acquisition module, a processing terminal, a field environment sensing module, and a data storage and transmission module. The homogeneous data acquisition module is fixed on the upper part of the mobile data acquisition frame and includes a hyperspectral imaging unit and a three-dimensional laser point cloud reconstruction unit that are deployed in conjunction with each other. The two units share the same spatial calibration reference. The field environment sensing module is electrically connected to the processing terminal. The processing terminal has a built-in fusion segmentation algorithm module, a three-layer index modeling calculation module, and a breeding grading and screening module. The mobile data acquisition frame is a tracked field mobile frame that is adapted to autonomous movement between rows in ridge-planted and flat-planted soybean fields.

Images