Intelligent crop seed vigor detection method and system

Abstract

The invention provides an intelligent crop seed vigor detection method and system. The method comprises the following steps: collecting and preprocessing seed data; extracting appearance, spectrum, hardness and internal quality characteristics of the seeds to generate a characteristic set; and inputting the feature set to the vigor prediction model to obtain a vigor prediction value, and dynamically adjusting a classification threshold to classify seed vigor. According to the method, quadruple features are integrated to construct a high-dimensional feature set, and the physiological activity and mechanical integrity of the seeds are accurately represented through a nanoindenter and an X-ray CT technology. An error feedback and optimization method is introduced, data changes are responded in real time, and the classification accuracy is improved. Micron-scale quantification is realized through X-ray CT tomography, and the embryo development evaluation accuracy is improved. The time sequence neural network is constructed, the dynamic relation between the characteristics is captured, the seed vigor descending trend is predicted in advance, and the long-term prediction stability is better.

Description

Technical Field

[0001] The present invention relates to the technical field of seed vitality detection, and in particular to an intelligent crop seed vitality detection method and system. Background Art

[0002] In agricultural production, seed vitality directly affects crop yield and quality, so rapid and accurate seed vitality detection is a key link in ensuring agricultural benefits. Traditional detection methods mainly rely on manual observation, germination tests or physical and chemical analysis of a single indicator, which has the problems of low efficiency, strong subjectivity, high destructiveness and difficulty in comprehensively evaluating the multi-dimensional characteristics of seeds. For example, conventional hardness tests use mechanical crushing methods, which can easily damage seeds and can only provide local data; internal quality inspections mostly rely on anatomical observations or low-resolution imaging, and cannot obtain three-dimensional structural information without loss; vitality prediction models are often based on a single spectral or morphological feature, ignoring the correlation between multimodal data, resulting in limited prediction accuracy. In addition, the classification thresholds mostly use fixed thresholds, which are difficult to adapt to the dynamic characteristics of seeds of different varieties and batches, and are prone to misjudgment due to environmental or sample differences.

[0003] In recent years, with the development of intelligent detection technology, some studies have attempted to improve detection accuracy through multi-source data fusion. For example, spectral analysis technology can reflect the chemical composition of seeds, but it is limited to single wavelength information; although X-ray imaging can obtain internal morphology, two-dimensional images are difficult to quantify the degree of embryo development; machine learning models have made certain progress in feature fusion, but static threshold classification still cannot cope with the problem of data drift. Existing technologies lack a systematic solution for comprehensive evaluation of seed vitality from multiple dimensions such as spectrum, morphology, mechanics, and internal structure, and have not solved the technical bottleneck of adaptive adjustment of dynamic classification thresholds, resulting in low stability and accuracy of detection results. Summary of the invention

[0004] The present invention aims to at least solve the technical problem of low stability and accuracy of detection results in the prior art, and particularly innovatively proposes an intelligent crop seed vitality detection method and system.

[0005] In order to achieve the above-mentioned object of the present invention, the present invention provides an intelligent crop seed vitality detection method, the method comprising: S1, collecting seed data and preprocessing the seed data; the seed data includes seed spectrum data, seed 360-degree image data, seed hardness data and seed internal quality image; S2, extracting features from the seed data to obtain appearance features, spectral features, hardness features, and internal quality features of the seeds, and generating a seed feature set; S3, inputting the feature set into a trained vitality prediction model to obtain a vitality prediction value; S4. Dynamically adjust the classification threshold, and classify the vitality of the seeds based on the vitality prediction value.

[0006] As an optional embodiment of the present invention, optionally, collecting seed hardness data in step S1 includes: S101. Select seed samples, remove surface impurities, and store them in a constant temperature box for more than 24 hours; S102, using a nanoindenter and an indenter, and setting parameters, selecting a flat area of ​​the seed without cracks or scratches to perform hardness tests at different points for at least three times, and obtaining elastic recovery data for each point; S103, collecting the elastic recovery data to obtain seed hardness data.

[0007] As an optional embodiment of the present invention, optionally, collecting the seed internal quality image in step S1 includes: S106, selecting seed samples of uniform size, removing surface impurities, and storing them in a constant temperature box for more than 24 hours; S107, placing the seed sample on a polytetrafluoroethylene tray with a rotating function, and performing a tomographic scan on the seed using an X-ray CT scanning device to obtain an internal tomographic image of the seed; S108, performing image enhancement processing on the internal tomographic image; S109, performing three-dimensional reconstruction on the enhanced internal tomographic image to obtain a three-dimensional image of the internal quality of the seed.

[0008] As an optional embodiment of the present invention, optionally, dynamically adjusting the classification threshold in step S4 includes: S401, determining an initial classification threshold according to historical classification accuracy; S402, after each classification, calculate the current classification accuracy, if the current classification accuracy is higher than the historical classification accuracy, keep the current classification threshold unchanged; S403: If the current classification accuracy is lower than the historical classification accuracy, the classification threshold is dynamically adjusted according to the classification error until the current classification accuracy is higher than or equal to the historical classification accuracy, or the preset upper limit of the number of adjustments is reached.

[0009] As an optional embodiment of the present invention, optionally, in step S403, if the current classification accuracy is lower than the historical classification accuracy, dynamically adjusting the classification threshold according to the classification error includes: S4031. If the current classification accuracy fluctuates in the short term, the classification threshold is adjusted using the error feedback method; S4032. If the current classification accuracy fluctuates in the long term, recalibrate the classification threshold in combination with the ROC curve optimization method; S4033. If the current classification accuracy is abnormal, switch to the adaptive window method and dynamically adjust the classification threshold; S4034: Verify the adjusted classification threshold.

[0010] On the other hand, the present invention also provides an intelligent crop seed vitality detection system, the system comprising an intelligent crop seed vitality detection method; the system further comprises: A collection module, used for collecting seed data of seeds; A processing module, connected to the acquisition module, for preprocessing and feature extraction of the collected seed data; A prediction module, connected to the processing module, for obtaining a predicted value of seed vitality using a vitality prediction model according to the extracted features; The classification module is connected to the prediction module and is used to classify seeds according to the vitality prediction value and dynamically adjust the classification threshold according to the historical classification accuracy.

[0011] Beneficial effects of the present invention: The present invention can integrate the four features of spectrum, morphology, mechanics, and internal structure to construct a high-dimensional feature set. For example, mechanical parameters such as elastic modulus and hardness are obtained by nanoindentation, and the embryo volume and porosity data of X-ray CT three-dimensional reconstruction can be combined to accurately characterize the physiological activity and mechanical integrity of seeds. The present invention introduces error feedback, ROC curve optimization and adaptive window method to respond to changes in data distribution in real time. When there are differences in seed batches or fluctuations in storage environment, dynamic threshold adjustment reduces the fluctuation range of classification accuracy from ±8% to ±3%. For example, in a sudden low temperature stress experiment, the dynamic threshold reduces the vitality classification error by 45%, effectively improving the stability of the inspection results. The present invention realizes micron-level quantification of parameters such as embryo volume and porosity through X-ray CT tomography and three-dimensional reconstruction. Compared with traditional two-dimensional imaging, the accuracy of embryo development assessment is improved, and dysplastic seeds (such as samples with embryo volume accounting for <75%) can be accurately identified, significantly improving the accuracy of vitality detection. A temporal neural network with a context layer is constructed to capture the dynamic coupling relationship between features. For example, through the correlation analysis of the attenuation law of spectral wavelength over time and hardness parameters, the downward trend of seed vitality can be predicted in advance, which has better long-term prediction stability.

[0012] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which: Figure 1This is a flow chart of an intelligent crop seed vitality detection method according to Embodiment 1 of the present invention; Figure 2 It is a structural schematic diagram of an intelligent crop seed vitality detection system according to embodiment 2 of the present invention. DETAILED DESCRIPTION

[0014] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0015] Example 1 like Figure 1 As shown, an intelligent method for detecting the vitality of crop seeds comprises: S1, collecting seed data and preprocessing the seed data; the seed data includes seed spectrum data, seed 360-degree image data, seed hardness data and seed internal quality image; It should be noted that when collecting seed spectral data, a portable near-infrared spectrometer (such as the ASDFieldSpec4 series) is used, with a wavelength range of 900-1700nm (suitable for seed component analysis), a spectral resolution of ≤3nm, and a signal-to-noise ratio of ≥1000:1. Equipped with a reflection probe (including a built-in whiteboard calibration module), a fiber optic transmission unit, and a sample cup holder. The seeds are vertically embedded in the fixture groove, and the position is adjusted so that the seed equatorial plane is parallel to the plane of the rotating table. The rotating table is set to rotate in steps of 10° (a total of 36 angles), and each angle triggers the camera to shoot. The light source intensity is adjusted to uniform brightness on the seed surface (grayscale value fluctuation <5%). Oversampling technology (30% overlap of each image) is used to ensure texture continuity. Multi-view images are stitched into a panoramic image through feature point matching (SIFT algorithm), and the resolution is normalized to 1024×1024 pixels. Blurred images are automatically removed (based on Laplacian operator clarity score, threshold ≥0.8). 10% of the stitching results are manually reviewed to ensure that the embryo area is not blocked or deformed.

[0016] When collecting 360-degree image data of seeds, an industrial linear array camera (resolution ≥5000×3000, frame rate ≥80fps) with a ring light source (color temperature 5500K, CRI ≥90) and a rotating stage (division value 0.01°, repeatability accuracy ±0.05°) was used for 360-degree rotation control of seeds. The background was black to eliminate environmental reflection interference, and a customized seed fixture (3D printing, made of matte plastic) was used to ensure that the geometric center of the seed coincided with the rotation axis. The seeds were vertically embedded in the fixture groove, and the position was adjusted so that the seed equatorial plane was parallel to the rotating stage plane. The rotating stage was set to rotate in steps of 10° (a total of 36 angles), and each angle triggered the camera to shoot. The light source intensity was adjusted to uniform brightness on the seed surface (grayscale value fluctuation <5%). Oversampling technology (30% overlap of each image) was used to ensure texture continuity. Multi-view images were stitched into a panorama by feature point matching (SIFT algorithm), and the resolution was normalized to 1024×1024 pixels. 10% of the stitching results were manually reviewed to ensure that there was no obstruction or deformation in the embryo area.

[0017] Preprocessing of seed data includes data cleaning, removing outliers and missing values ​​to ensure data integrity and accuracy. For spectral data, smoothing filtering is performed to reduce noise interference; for image data, image enhancement and denoising are performed to improve image quality. In addition, seed data needs to be standardized to convert data of different dimensions to the same scale to facilitate subsequent feature extraction and analysis.

[0018] S2, extracting features from the seed data to obtain appearance features, spectral features, hardness features, and internal quality features of the seeds, and generating a seed feature set; It should be noted that in this embodiment, the appearance features are extracted by using a deep learning algorithm, such as a convolutional neural network (CNN), to process the 360-degree image data of the seeds, automatically identify and extract the appearance features of the seeds such as shape, color, and texture. The spectral data of the seeds is obtained by using a near-infrared spectrometer, and the dimensionality is reduced by methods such as principal component analysis (PCA) to extract key spectral features, which can reflect the chemical composition and physiological state of the seeds.

[0019] S3, inputting the feature set into a trained vitality prediction model to obtain a vitality prediction value; It should be noted that the vitality prediction model of this embodiment adopts a time-aware neural network (Time-Aware Neural Network) with a context layer, which is specially designed to process multimodal time series features of seeds. The input layer receives the fused four-dimensional feature vector (spectral, morphological, mechanical, internal structure) and inputs it after standardization. The hidden layer contains a bidirectional LSTM unit to capture the dynamic correlation between features (such as the coupling effect of spectral changes and hardness attenuation). The context layer stores historical hidden states to enhance the model's sensitivity to long-term dependencies. The output layer, a fully connected layer, outputs the seed vitality value (range 0-1, the higher the value, the stronger the vitality).

[0020] When training the vitality prediction model: The training data uses data from multiple batches of seed life cycle data (including labels such as germination rate and storage time). Stratified sampling is used to ensure that samples of different vitality levels are balanced. The spectral reflectance curve, 360° image texture features, nanoindentation mechanical parameters, and CT three-dimensional embryo structure data are mapped to the same dimension space through maximum and minimum value normalization. The attention mechanism (Self-Attention) is used to dynamically weight key features (such as the spectral red edge displacement during the critical period of germination). Seeds are regularly inspected during storage, and a time series feature matrix is ​​constructed to capture the vitality decay trend.

[0021] The process of generating predicted values ​​is as follows: The input feature vector passes through the LSTM unit in sequence, and the context layer updates the historical state. The final hidden layer state is mapped to the vitality prediction value through the fully connected layer. Monte Carlo Dropout is used to perform multiple forward propagations in the test phase to generate the distribution interval of the prediction value and evaluate the model confidence.

[0022] Model validation and optimization are: Combine the mean square error (MSE) with the smooth L1 loss to balance outlier robustness and gradient stability. Use cosine annealing learning rate scheduling, with an initial learning rate of 0.001 and a batch size of 128. Early Stopping is used to prevent overfitting, and training is terminated if the validation set loss does not decrease for 5 consecutive rounds.

[0023] The coefficient of determination (R²) calculated on the independent test set was >0.92, and the correlation coefficient between the predicted value and the true germination rate was >0.95. The classification accuracy of the vitality level was analyzed through the confusion matrix to ensure that the recognition rate of high-vigor seeds (>0.8) reached more than 98%. The network weights were fine-tuned regularly (e.g., quarterly) with new batches of data to adapt to seasonal changes in seed physiological characteristics. The distribution deviation of the predicted values ​​was monitored, and if the confidence of the continuous sample prediction was lower than the threshold, the model retraining process was triggered.

[0024] Through the above process, the model can robustly output high-precision vitality prediction values, and combined with the dynamic threshold classification strategy, it can significantly improve the reliability and practicality of seed vitality assessment.

[0025] S4. Dynamically adjust the classification threshold, and classify the vitality of the seeds based on the vitality prediction value.

[0026] It should be noted that in step S4, the seed vitality level threshold is divided into three types: first-level seeds: predicted value ≥ 0.85 (high vitality, suitable for long-term storage or precision sowing); second-level seeds: 0.7≤ predicted value < 0.85 (medium vitality, need short-term storage or for ordinary planting); third-level seeds: predicted value < 0.7 (low vitality, recommended to be eliminated or used for scientific research experiments); In the process of dynamically adjusting the classification threshold, the system will make intelligent judgments based on the historical classification accuracy and the characteristics of the current batch of seeds. For example, when the system detects that the overall vitality level of the current batch of seeds is high, it will automatically increase the classification threshold to ensure that high-vigor seeds can be accurately identified and retained. On the contrary, if the system detects that the seed vitality is generally low, it will lower the classification threshold accordingly to avoid mistakenly classifying potential seeds as low-vigor seeds. This dynamic adjustment mechanism enables the seed vitality detection system of the present invention to adapt to the characteristics of different batches of seeds and improve the accuracy and practicality of classification.

[0027] In summary, the intelligent crop seed vitality detection method of this embodiment can integrate the four features of spectrum, morphology, mechanics, and internal structure to construct a high-dimensional feature set. For example, mechanical parameters such as elastic modulus and hardness are obtained by nanoindentation, and the embryo volume and porosity data of X-ray CT three-dimensional reconstruction can be combined to accurately characterize the physiological activity and mechanical integrity of seeds. The present invention introduces error feedback, ROC curve optimization and adaptive window method to respond to data distribution changes in real time. When there are differences in seed batches or fluctuations in storage environment, dynamic threshold adjustment reduces the fluctuation range of classification accuracy from ±8% to ±3%. For example, in a sudden low temperature stress experiment, the dynamic threshold reduces the vitality classification error by 45%, effectively improving the stability of the inspection results. The present invention realizes micron-level quantification of parameters such as embryo volume and porosity through X-ray CT tomography and three-dimensional reconstruction. Compared with traditional two-dimensional imaging, the accuracy of embryo development assessment is improved, and dysplastic seeds (such as samples with embryo volume accounting for <75%) can be accurately identified, significantly improving the accuracy of vitality detection. A temporal neural network with a context layer is constructed to capture the dynamic coupling relationship between features. For example, by analyzing the correlation between the decay of spectral wavelength over time and hardness parameters, the decline in seed vitality can be predicted in advance. Actual measurements show that the accuracy of this model in identifying aged seeds is 18% higher than that of the static model, and it has better long-term prediction stability.

[0028] As an optional embodiment of the present invention, optionally, collecting seed hardness data in step S1 includes: S101. Select seed samples, remove surface impurities, and store them in a constant temperature box for more than 24 hours; It should be noted that when selecting seed samples in step S101, it is necessary to ensure that the samples are representative and can reflect the overall characteristics of the batch of seeds. The purpose of removing surface impurities is to avoid the interference of impurities on the hardness measurement results. The seed samples are stored in a constant temperature box for more than 24 hours to allow the seeds to reach a temperature equilibrium state and eliminate the influence of temperature on the hardness measurement results.

[0029] S102, using a nanoindenter and an indenter, and setting parameters, selecting a flat area of ​​the seed without cracks or scratches to perform hardness tests at different points for at least three times, and obtaining elastic recovery data for each point; It should be noted that when using a nano indenter to perform hardness testing in step S102, it is necessary to ensure that the indenter is clean and wear-free to avoid errors in the test results. Parameters such as indenter penetration depth and loading rate are set, and the selection of these parameters should be reasonably set according to the characteristics of the seed and the test requirements. A flat area without cracks or scratches on the seed surface is selected for testing to ensure the accuracy and reliability of the test results. At least three hardness tests are performed on each selected point in order to obtain a more accurate average hardness value and reduce the error that may be caused by a single test. By recording the elastic recovery data of each point, the mechanical properties and elastic recovery of the seed can be further analyzed, providing more comprehensive data support for the vitality assessment of the seed.

[0030] S103, collecting the elastic recovery data to obtain seed hardness data.

[0031] It should be noted that in step S103, the elastic recovery data obtained from multiple hardness tests are processed collectively to calculate the average hardness value to reflect the overall hardness characteristics of the seeds. As an important component of the internal quality characteristics of the seeds, the hardness data can reflect the mechanical strength and tolerance of the seeds and provide key information for subsequent vitality prediction. By accurately collecting and processing the hardness data, the accuracy and reliability of seed vitality detection can be further improved.

[0032] As an optional embodiment of the present invention, optionally, the expression for calculating the hardness characteristic in step S2 is: , , , ; in, Indicates hardness characteristics, Represents a splicing operation, represents the elastic modulus, It indicates hardness. represents the resonant frequency, represents the unloading stiffness, Indicates the geometric correction factor of the pressure head, represents the contact area, Indicates the maximum indentation load, represents the equivalent stiffness, Indicates seed quality.

[0033] As an optional embodiment of the present invention, optionally, collecting the seed internal quality image in step S1 includes: S106, selecting seed samples of uniform size, removing surface impurities, and storing them in a constant temperature box for more than 24 hours; S107, placing the seed sample on a polytetrafluoroethylene tray with a rotating function, and performing a tomographic scan on the seed using an X-ray CT scanning device to obtain an internal tomographic image of the seed; It should be noted that when the seed is tomographically scanned by an X-ray CT scanning device in step S107, it is necessary to ensure that the scanning parameters are reasonably set, such as tube voltage, tube current, scanning speed, etc., to obtain a clear internal tomographic image. The seed sample is placed on a polytetrafluoroethylene tray with a rotating function to ensure that the seed can rotate evenly during the scanning process, thereby obtaining a full range of internal image information. The seed internal tomographic image obtained by tomography can clearly show the internal structure of the seed, such as the position, shape, size and internal pore distribution of the embryo, etc., providing an important basis for subsequent internal quality feature extraction. After obtaining the internal tomographic image, image reconstruction and processing are also required to improve image quality and resolution, so as to facilitate subsequent feature extraction and analysis.

[0034] S108, performing image enhancement processing on the internal tomographic image; It should be noted that in step S108, the internal tomographic image is subjected to image enhancement processing, aiming to improve the visualization effect and detail expression of the image, and provide clearer and more accurate image information for subsequent feature extraction and analysis. Image enhancement processing may include operations such as contrast adjustment, noise suppression, and edge sharpening, which may be reasonably selected and combined according to the specific conditions and requirements of the image. By contrast adjustment, the brightness difference between different objects or regions in the image may be increased, making the image clearer and easier to observe. Noise suppression may remove random noise and interference in the image, and improve the purity and signal-to-noise ratio of the image. Edge sharpening may enhance the edges and contours of objects in the image, making the image more three-dimensional and vivid.

[0035] S109, performing three-dimensional reconstruction on the enhanced internal tomographic image to obtain a three-dimensional image of the internal quality of the seed.

[0036] It should be noted that in step S109, the internal tomographic image after the enhanced processing is carried out three-dimensional reconstruction. Through advanced image reconstruction algorithm, two-dimensional tomographic images are superimposed and merged to form the internal three-dimensional structure of seed. This step not only requires a high degree of image processing capability, but also needs to have a deep understanding of the internal structure of seed to ensure the accuracy and authenticity of the reconstruction result. The reconstructed three-dimensional image can intuitively show the internal structure of seed, such as the three-dimensional morphology of embryo, internal pore network, etc., for subsequent feature extraction provides more intuitive and comprehensive information. After obtaining the internal quality three-dimensional image, further analysis and processing are required to extract the key features relevant to seed vitality.

[0037] As an optional embodiment of the present invention, optionally, the expression for extracting the internal quality feature in step S2 is: , , , ; in, Represents internal quality characteristics, Represents a splicing operation, represents the internal porosity of the seed, represents the actual embryo volume, represents the theoretical maximum embryo volume, represents the sound wave attenuation coefficient, represents the propagation distance, represents the logarithmic function, represents the initial amplitude, Indicates the received amplitude, represents the proportion of embryo area, represents the pixel area of ​​the embryo region, Represents the total pixel area of ​​the seed.

[0038] As an optional embodiment of the present invention, optionally, in step S3, the expression for obtaining the vitality prediction value using the vitality prediction model is: , , , ; in, Indicates that at time step When the output layer The predicted output value of each neuron, represents the total number of neurons, represents the hidden layer neurons to the output layer The connection weights of neurons, represents the hidden layer The neuron at time step The output, represents the activation function, Represents the hidden layer neurons The net input, represents the input feature dimension, Represents the input layer neurons to the hidden layer The connection weights of neurons, Represents the input layer The neuron at time step The input value of Represents the context layer neurons to the hidden layer The feedback weight of each neuron, Represents the context layer The neuron at the previous time step The feedback value of Indicates that at time step After that, the context layer neurons The state of the current hidden layer neuron is updated The output of is used as feedback for the next time step.

[0039] As an optional embodiment of the present invention, optionally, dynamically adjusting the classification threshold in step S4 includes: S401, determining an initial classification threshold according to historical classification accuracy; It should be noted that the setting of the initial classification threshold in step S401 is based on a large amount of historical data and classification results, and is obtained through statistical analysis. A reasonable initial classification threshold can improve the accuracy and stability of classification. The setting of the initial classification threshold needs to consider multiple factors, such as seed type, storage conditions, test environment, etc., to ensure its adaptability and accuracy.

[0040] S402, after each classification, calculate the current classification accuracy, if the current classification accuracy is higher than the historical classification accuracy, keep the current classification threshold unchanged; It should be noted that, in step S402, by dynamically adjusting the classification threshold, the classification result can be continuously optimized to make it more in line with the actual situation. If the current classification accuracy is lower than the historical classification accuracy, it means that the current classification threshold may not be suitable for the current test environment or seed characteristics, and the classification threshold needs to be adjusted at this time. The adjustment method can be gradual fine-tuning, or it can be adjusted according to specific rules or algorithms. In the adjustment process, it is necessary to continuously monitor the changes in the classification accuracy to ensure the effectiveness and accuracy of the adjustment. Through continuous adjustment and optimization, the accuracy and stability of the classification can be gradually improved.

[0041] S403: If the current classification accuracy is lower than the historical classification accuracy, the classification threshold is dynamically adjusted according to the classification error until the current classification accuracy is higher than or equal to the historical classification accuracy, or the preset upper limit of the number of adjustments is reached.

[0042] It should be noted that in step S403, when the classification accuracy is continuously lower than the historical classification accuracy, the system will automatically adjust the classification threshold according to the size and direction of the classification error. This adjustment is dynamic and intelligent, and is intended to gradually reduce the classification error and improve the accuracy and stability of the classification. During the adjustment process, the system will take into account a variety of factors, such as the distribution of the classification error, the change of seed characteristics, etc., to ensure the rationality and effectiveness of the adjustment. If after multiple adjustments, the current classification accuracy is still lower than the historical classification accuracy, and has reached the preset upper limit of the number of adjustments, the system will prompt the user to recheck the test environment, seed samples or classification models, etc., to find out the reasons that may cause the classification accuracy to decline, and make corresponding improvements and optimizations. Through this series of dynamic adjustments and optimizations, the intelligent crop seed vitality detection method and system of the present invention can achieve accurate and rapid evaluation of seed vitality.

[0043] As an optional embodiment of the present invention, optionally, in step S403, if the current classification accuracy is lower than the historical classification accuracy, dynamically adjusting the classification threshold according to the classification error includes: S4031. If the current classification accuracy fluctuates in the short term, the classification threshold is adjusted using the error feedback method; It should be noted that in step S4031, the error feedback method is a feedback adjustment mechanism based on classification error. When the system detects that the current classification accuracy fluctuates in the short term, it will be considered that this is caused by random factors or short-term environmental changes. At this time, the error feedback method is used for adjustment. The error feedback method will fine-tune the classification threshold according to the size and direction of the classification error to quickly respond to such short-term fluctuations and quickly restore the classification accuracy to a normal level. Through the adjustment of the error feedback method, the system can maintain the stability and accuracy of the classification results and avoid the accumulation of classification errors caused by short-term fluctuations.

[0044] S4032. If the current classification accuracy fluctuates in the long term, recalibrate the classification threshold in combination with the ROC curve optimization method; It should be noted that in step S4032, if the current classification accuracy shows a fluctuation trend in the long term, it may mean that the test environment, seed characteristics or the classification model itself has undergone a more significant change. At this time, simply relying on the error feedback method for adjustment may not achieve the desired effect. Therefore, the present invention adopts a strategy of recalibrating the classification threshold in combination with the ROC curve optimization method. The ROC curve (Receiver Operating Characteristic Curve) is a tool for evaluating the performance of a classification model. By drawing the relationship curve between the true positive rate (True Positive Rate) and the false positive rate (False Positive Rate) under different classification thresholds, the classification performance of the model can be intuitively displayed. Combined with the ROC curve optimization method, the system can analyze the distribution of classification errors and find the optimal classification threshold, thereby achieving recalibration of the classification results. Through this strategy, the present invention can cope with long-term fluctuations in classification accuracy and ensure the accuracy and stability of the classification results. After recalibrating the classification threshold, the system will continue to monitor changes in the classification accuracy and make further adjustments and optimizations as needed.

[0045] S4033. If the current classification accuracy is abnormal, switch to the adaptive window method and dynamically adjust the classification threshold; It should be noted that in step S4033, if the current classification accuracy rate has a sudden abnormality, this may be due to an abnormal situation in the test environment, such as extreme weather, equipment failure, etc., or there are special problems in the seed sample itself. At this time, the system needs to respond quickly and switch to the adaptive window method for adjustment. The adaptive window method is a dynamic adjustment strategy based on a sliding window. It automatically adjusts the size and position of the window according to the changing trend and fluctuation range of the classification accuracy rate to achieve dynamic adjustment of the classification threshold. Through the adjustment of the adaptive window method, the system can quickly adapt to the changes in the test environment, reduce classification errors, and improve the accuracy and stability of classification. During the adjustment process, the system will continue to monitor the changes in the classification accuracy rate, and make further adjustments and optimizations as needed. If the classification accuracy rate still cannot be restored to a normal level after the adjustment of the adaptive window method, the system will prompt the user to make further inspections and adjustments to ensure the normal operation of the system and the accuracy of the classification results.

[0046] S4034: Verify the adjusted classification threshold.

[0047] It should be noted that verifying the adjusted classification threshold in step S4034 is a key step to ensure the effect of the adjustment. The verification process mainly includes two parts: one is to test the adjusted classification threshold using the verification data set to evaluate its classification performance; the other is to compare the verification result with the historical classification accuracy to confirm whether the adjustment is effective.

[0048] During the verification process, the system will use methods such as cross-validation to ensure the reliability and stability of the verification results. By comparing the verification results with the historical classification accuracy, the system can determine whether the adjusted classification threshold has achieved the expected effect. If the verification results show that the adjusted classification threshold can significantly improve the classification accuracy, the system will formally adopt the classification threshold for subsequent seed vitality evaluation. If the verification results are not ideal, the system will continue to adjust the classification threshold until the optimal solution is found.

[0049] As an optional embodiment of the present invention, optionally, in step S4031, the expression for adjusting the classification threshold using the error feedback method is: ; in, represents the updated classification threshold, It indicates the current classification threshold. represents the learning rate, represents the number of samples, Indicates The true labels of samples, Indicates The predicted labels of samples, Indicates the direction of sensitivity of the predicted value to the threshold; In step S4032, the expression for recalibrating the classification threshold in combination with the ROC curve optimization method is: ; in, represents the optimal classification threshold, Indicates that the threshold The area under the ROC curve; The expression for dynamically adjusting the classification threshold in step S4033 is: , ; in, represents the mixed classification threshold, represents the smoothing coefficient, represents the dynamic classification threshold, represents the historical optimal threshold, Represents a sliding window The mean of the internal prediction probabilities, represents the adjustment coefficient, Represents a sliding window The standard deviation of the predicted probabilities.

[0050] Example 2 like Figure 2 As shown, an intelligent crop seed vitality detection system, the system includes an intelligent crop seed vitality detection method; the system is used to implement the intelligent crop seed vitality detection method; the system also includes: A collection module, used for collecting seed data of seeds; It should be noted that the collection module is mainly responsible for collecting various data related to seed vitality assessment, including but not limited to key indicators such as seed appearance characteristics, size, weight, germination rate, and growth rate.

[0051] A processing module, connected to the acquisition module, for preprocessing and feature extraction of the collected seed data; It should be noted that the processing module is mainly responsible for cleaning, screening and feature extraction of the collected seed data. By preprocessing the raw data, noise and outliers can be removed to improve the accuracy and reliability of the data. Feature extraction is to extract key features that have an important impact on seed vitality assessment from the raw data. The processing module uses advanced algorithms and technologies to efficiently process large amounts of data, ensuring the real-time and accuracy of the system.

[0052] A prediction module, connected to the processing module, for obtaining a predicted value of seed vitality using a vitality prediction model according to the extracted features; It should be noted that the prediction module uses a pre-trained vitality prediction model to analyze and calculate the features extracted by the processing module to obtain the predicted vitality value of the seeds. This predicted value can reflect the growth potential and germination ability of the seeds, providing an important reference for subsequent seed screening and planting. The vitality prediction model used by the prediction module has been trained and verified with a large amount of data, has a high degree of accuracy and stability, and can cope with seeds of different types and characteristics.

[0053] The classification module is connected to the prediction module and is used to classify seeds according to the vitality prediction value and dynamically adjust the classification threshold according to the historical classification accuracy.

[0054] It should be noted that the classification module is mainly responsible for classifying the vitality prediction values ​​output by the prediction module to determine the vitality level of the seeds. The classification process is based on preset classification standards and thresholds. At the same time, the classification module also has the ability to dynamically adjust the classification threshold, and can automatically adjust the classification threshold according to the historical classification accuracy and the current classification results to ensure the accuracy and stability of the classification. Through continuous adjustment and optimization, the classification module can adapt to seeds of different types and characteristics and achieve accurate assessment of seed vitality. The entire intelligent crop seed vitality detection system can achieve rapid and accurate assessment of seed vitality through the collaborative work of various modules, providing strong technical support for agricultural production.

[0055] The principle of the intelligent crop seed vitality detection system is as follows: the collection module collects various types of seed data, the processing module preprocesses and extracts features from the data, the prediction module uses the vitality prediction model to obtain the predicted vitality value of the seeds, and finally the classification module classifies the seeds according to the predicted vitality value. During the classification process, the system not only relies on the preset classification standards and thresholds, but also has the ability to intelligently adjust the classification thresholds. This dynamic adjustment mechanism ensures the accuracy and stability of the classification, enabling the system to adapt to seeds of different types and characteristics. Through the collaborative work of the entire system, a comprehensive, rapid and accurate evaluation of seed vitality is achieved.

[0056] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. An intelligent method for detecting the vitality of crop seeds, characterized in that: The method comprises: S1, collecting seed data and preprocessing the seed data; the seed data includes seed spectrum data, seed 360-degree image data, seed hardness data and seed internal quality image; S2, extracting features from the seed data to obtain appearance features, spectral features, hardness features, and internal quality features of the seeds, and generating a seed feature set; S3, inputting the feature set into a trained vitality prediction model to obtain a vitality prediction value; S4. Dynamically adjust the classification threshold, and classify the vitality of the seeds based on the vitality prediction value.

2. The intelligent crop seed vitality detection method according to claim 1, characterized in that: Collecting seed hardness data in step S1 includes: S101. Select seed samples, remove surface impurities, and store them in a constant temperature box for more than 24 hours; S102, using a nanoindenter and an indenter, and setting parameters, selecting a flat area of ​​the seed without cracks or scratches to perform hardness tests at different points for at least three times, and obtaining elastic recovery data for each point; S103, collecting the elastic recovery data to obtain seed hardness data.

3. The intelligent crop seed vitality detection method according to claim 2, characterized in that: The expression for calculating the hardness characteristic in step S2 is: ， ， ， ； in, Indicates hardness characteristics, Represents a splicing operation, represents the elastic modulus, It indicates hardness. represents the resonant frequency, represents the unloading stiffness, represents the geometric correction factor of the pressure head, represents the contact area, Indicates the maximum indentation load, represents the equivalent stiffness, Indicates seed quality.

4. The intelligent crop seed vitality detection method according to claim 1, characterized in that: In step S1, collecting the seed internal quality image includes: S106, selecting seed samples of uniform size, removing surface impurities, and storing them in a constant temperature box for more than 24 hours; S107, placing the seed sample on a polytetrafluoroethylene tray with a rotating function, and performing a tomographic scan on the seed using an X-ray CT scanning device to obtain an internal tomographic image of the seed; S108, performing image enhancement processing on the internal tomographic image; S109, performing three-dimensional reconstruction on the enhanced internal tomographic image to obtain a three-dimensional image of the internal quality of the seed.

5. The intelligent crop seed vitality detection method according to claim 4, characterized in that: The expression for extracting the internal quality features in step S2 is: ， ， ， ； in, Represents internal quality characteristics, Represents a splicing operation, represents the internal porosity of the seed, represents the actual embryo volume, represents the theoretical maximum embryo volume, represents the sound wave attenuation coefficient, represents the propagation distance, represents the logarithmic function, represents the initial amplitude, Indicates the received amplitude, represents the proportion of embryo area, represents the pixel area of ​​the embryo region, Represents the total pixel area of ​​the seed.

6. The intelligent crop seed vitality detection method according to claim 1, characterized in that: In step S3, the expression for obtaining the predicted vitality value using the vitality prediction model is: ， ， ， ； in, Indicates that at time step When the output layer The predicted output value of each neuron, represents the total number of neurons, represents the hidden layer neurons to the output layer The connection weights of neurons, represents the hidden layer The neuron at time step The output, represents the activation function, Represents the hidden layer neurons The net input, represents the input feature dimension, Represents the input layer neurons to the hidden layer The connection weights of neurons, Represents the input layer The neuron at time step The input value of Represents the context layer neurons to the hidden layer The feedback weight of each neuron, Represents the context layer The neuron at the previous time step The feedback value of Indicates that at time step After that, the context layer neurons The state of the current hidden layer neuron is updated The output of is used as feedback for the next time step.

7. The intelligent crop seed vitality detection method according to claim 1, characterized in that: Dynamically adjusting the classification threshold in step S4 includes: S401, determining an initial classification threshold according to historical classification accuracy; S402, after each classification, calculate the current classification accuracy, if the current classification accuracy is higher than the historical classification accuracy, keep the current classification threshold unchanged; S403: If the current classification accuracy is lower than the historical classification accuracy, the classification threshold is dynamically adjusted according to the classification error until the current classification accuracy is higher than or equal to the historical classification accuracy, or the preset upper limit of the number of adjustments is reached.

8. The intelligent crop seed vitality detection method according to claim 7, characterized in that: In step S403, if the current classification accuracy is lower than the historical classification accuracy, dynamically adjusting the classification threshold according to the classification error includes: S4031. If the current classification accuracy fluctuates in the short term, the classification threshold is adjusted using the error feedback method; S4032. If the current classification accuracy fluctuates in the long term, recalibrate the classification threshold in combination with the ROC curve optimization method; S4033. If the current classification accuracy is abnormal, switch to the adaptive window method and dynamically adjust the classification threshold; S4034: Verify the adjusted classification threshold.

9. The intelligent crop seed vitality detection method according to claim 8, characterized in that: The expression for adjusting the classification threshold using the error feedback method in step S4031 is: ； in, represents the updated classification threshold, It indicates the current classification threshold. represents the learning rate, represents the number of samples, Indicates The true labels of samples, Indicates The predicted labels of samples, Indicates the direction of sensitivity of the predicted value to the threshold; In step S4032, the expression for recalibrating the classification threshold in combination with the ROC curve optimization method is: ； in, represents the optimal classification threshold, Indicates that the threshold The area under the ROC curve; The expression for dynamically adjusting the classification threshold in step S4033 is: ， ； in, represents the mixed classification threshold, represents the smoothing coefficient, represents the dynamic classification threshold, represents the historical optimal threshold, Represents a sliding window The mean of the internal prediction probabilities, represents the adjustment coefficient, Represents a sliding window The standard deviation of the predicted probabilities.

10. An intelligent crop seed vitality detection system, characterized in that: The system includes the intelligent crop seed vitality detection method according to claims 1 to 9; the system also includes: A collection module, used for collecting seed data of seeds; A processing module, connected to the acquisition module, for preprocessing and feature extraction of the collected seed data; A prediction module, connected to the processing module, for obtaining a predicted value of seed vitality using a vitality prediction model according to the extracted features; The classification module is connected to the prediction module and is used to classify seeds according to the vitality prediction value and dynamically adjust the classification threshold according to the historical classification accuracy.

Images