Method for evaluating growth cycle of medicinal plant based on AI image recognition

Abstract

The present application relates to the field of intelligent evaluation and management of the growth cycle of medicinal plants, and provides a method for evaluating the growth cycle of medicinal plants based on AI image recognition, which comprises collecting multi-dimensional image sequences of the growth process of plants through a multispectral imaging device, synchronously recording ambient light intensity and temperature and humidity data; performing spatio-temporal registration on the image sequences to eliminate pixel offset caused by plant swinging and changes in shooting angle; using an adaptive hybrid filtering algorithm for denoising to preserve leaf edge texture details; performing multi-scale contrast enhancement on the denoised images to reconstruct high-frequency feature maps; constructing a double-branch deep network to extract leaf morphological dynamic feature sets and physiological parameter change feature sets, respectively; performing cross-modal fusion on the two feature sets to calculate a dynamic growth index matrix in combination with environmental parameters; and based on matching degree analysis of the dynamic growth index matrix and a preset growth stage model, outputting a current growth cycle stage identifier and a stage conversion probability curve.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent evaluation and management of medicinal plant growth cycle, and more particularly, to an AI image recognition-based medicinal plant growth cycle evaluation method. BACKGROUND

[0002] Evaluation and management of medicinal plant growth cycle is an important link in agricultural production and drug research and development. Traditional evaluation methods mainly rely on manual observation and experience judgment, which is difficult to realize precise monitoring and dynamic analysis of plant growth state. With the development of multispectral imaging technology and deep learning algorithm, plant growth cycle evaluation methods based on image recognition have gradually become a research hotspot. However, the existing technology still has many deficiencies in multi-dimensional image acquisition, feature extraction and growth stage division, etc. For example, the spatio-temporal registration accuracy of multispectral images is low, which is difficult to eliminate the influence of plant swing and shooting angle change; in the feature extraction process, the fusion effect of leaf morphological features and physiological parameter features is not good, resulting in insufficient accuracy of growth index matrix; in addition, the existing growth stage model has poor adaptability to complex environmental factors, and it is difficult to realize precise stage division and prediction.

[0003] In the implementation process of the embodiments of the present application, the inventors have found that the existing technology at least has the following problems or defects: insufficient accuracy of multi-dimensional image acquisition and registration, poor feature extraction and fusion effect, and poor adaptability of growth stage model to complex environment. SUMMARY

[0004] The present application provides an AI image recognition-based medicinal plant growth cycle evaluation method, comprising:

[0005] S1, acquiring a multi-dimensional image sequence of the growth process of medicinal plants by a multispectral imaging device, and synchronously recording environmental light intensity and temperature and humidity data;

[0006] S2, performing a spatio-temporal registration operation on the multi-dimensional image sequence to eliminate pixel shift caused by plant swing and shooting angle change;

[0007] S3, using an adaptive hybrid filtering algorithm to eliminate noise from the registered images, and retaining leaf edge texture details;

[0008] S4, performing multi-scale contrast enhancement on the denoised images, and reconstructing high-frequency feature maps through Laplacian pyramid decomposition;

[0009] S5, constructing a double-branch deep network, the first branch extracting a set of leaf morphological dynamic features, and the second branch extracting a set of physiological parameter change features;

[0010] S6, cross-modal fusion of the morphological dynamic feature set and the physiological parameter change feature set, calculation of a dynamic growth index matrix combined with an environmental parameter;

[0011] S7, based on the matching degree analysis of the dynamic growth index matrix and the preset growth stage model, output of a current growth cycle stage identifier and a stage conversion probability curve.

[0012] Further, the specific steps of extracting the leaf morphological dynamic feature set in step S5 include:

[0013] S51, performing a dynamic deformation perception convolution operation on the visible light band image sequence, the convolution kernel adaptively adjusting a spatial sampling position according to a leaf growth direction, and generating a leaf contour deformation field;

[0014] S52, calculation of a three-dimensional curvature integral value based on the deformation field, and establishment of a leaf curling degree;

[0015] S53, joint encoding of the leaf curling degree index and a leaf vein fractal dimension, and generation of the morphological dynamic feature set.

[0016] Further, the specific steps of extracting the physiological parameter change feature set in step S5 include:

[0017] S54, construction of a spectral reflectance dynamic decoupling model, separation of a chlorophyll absorption band and a water molecule frequency doubling vibration signal;

[0018] S55, calculation of a metabolic activity index according to a spectral fingerprint library of medicinal plant species;

[0019] S56, time domain cross-correlation analysis of the metabolic activity index and transpiration fluctuation rate, and generation of the physiological parameter change feature set.

[0020] Further, the cross-modal fusion in step S6 adopts:

[0021] S61, construction of a growth trend coupling model, mapping of the morphological dynamic feature set to a physiological parameter space, and calculation of a feature compatibility matrix;

[0022] S62, generation of a growth constraint factor based on an environmental parameter, the factor including a modulation coefficient of light intensity on photosynthesis phase and a damping coefficient of temperature and humidity on metabolic rate;

[0023] S63, execution of multi-source feature constraint optimization as shown in the following formula:

[0024]

[0025] wherein, is the fused dynamic growth index matrix; A set of morphological dynamic features; A set of features representing changes in physiological parameters; , For adaptive balance parameters; For environmental constraint functions; The growth index matrix to be optimized; It is a set of environmentally related factors.

[0026] Furthermore, the environmental constraint function The calculation includes:

[0027] S611. The phase modulation equation for photosynthetically active radiation is established as shown in the following formula:

[0028]

[0029] in, L is the illumination phase modulation factor; L is the real-time illumination intensity. The light saturation intensity of the species; The day-night cycle parameter is t; t is the current time.

[0030] S612. Construct the metabolic damping term as shown in the following formula:

[0031]

[0032] in, The damping factor is the temperature and humidity; T is the ambient temperature. The optimal growth temperature; H represents the temperature tolerance range; H represents the ambient humidity. This represents the critical humidity threshold.

[0033] Furthermore, the specific implementation of step S7 includes:

[0034] S71. Construct a quantized model of the growth stage, map the dynamic growth index matrix to Hilbert space, and generate discrete energy level basis vectors.

[0035] S72. Calculate the coherence between the current dynamic growth index matrix and the basis vectors of each energy level;

[0036] S73. Generate the stage transition probability curve based on the coherence, and use the Hilbert space basis vectors as topological constraints for the state transition path.

[0037] Furthermore, the calculation method for the stage transition probability curve includes:

[0038] S74. Calculate the barrier height between adjacent energy levels based on the Hilbert space basis vectors. Thermal fluctuation energy ;

[0039] S75. The quantum tunneling rate equation is established as shown in the following formula:

[0040]

[0041] in, This refers to the stage transition rate; This refers to the intrinsic transition rate; The height of the barrier; This refers to thermal fluctuation energy; is Boltzmann's constant; T is the ambient temperature.

[0042] Furthermore, the adaptive hybrid filtering algorithm in step S3 includes:

[0043] S31, Perform anisotropic diffusion filtering for blade guidance;

[0044] S32. Perform morphological wavelet packet decomposition in the frequency domain to retain sub-band components that match the vibration frequency of the leaf edge.

[0045] Furthermore, the dynamic deformation-aware convolution operation includes:

[0046] S511. Generate convolution kernel deformation offset based on the leaf growth direction field;

[0047] S512. Apply the offset to the convolution kernel coordinates through differentiable sampling to generate the deformed feature map.

[0048] Furthermore, the quantum model of the growth stages includes:

[0049] S711. Construct a growth Hilbert space whose basis vectors are composed of photosynthetic eigenstates;

[0050] S712. Introduce environmental parameter perturbation Hamiltonian to split the Hilbert space basis vector into energy levels;

[0051] S713. Solve for the eigenstates of the Hamiltonian using the imaginary time evolution algorithm to determine the stage transition path.

[0052] The embodiments of the present invention have at least the following beneficial effects: The present invention can improve the accuracy and efficiency of medicinal plant growth cycle assessment. By combining multispectral imaging technology with deep learning algorithms, changes in plant morphological and physiological parameters can be monitored in real time, thereby accurately identifying their growth stages. This method can avoid the subjectivity and errors of traditional manual observation, providing a reliable basis for the scientific management of medicinal plants. In addition, by combining multidimensional analysis of environmental parameters, the growth cycle model can be dynamically adjusted to adapt to the plant growth needs under different climatic and soil conditions, further improving the applicability and practicality of the assessment results.

[0053] This invention can also optimize the harvesting timing of medicinal plants and the drug development process. Through precise growth stage segmentation and dynamic growth index analysis, the optimal period for the accumulation of active ingredients in plants can be determined, thereby improving the quality and efficacy of medicinal materials. Simultaneously, based on the predictive function of the growth cycle model, planting and harvesting plans can be planned in advance, reducing resource waste and improving production efficiency. This technology can provide high-quality plant raw materials for drug development, accelerate the development of new drugs, and promote the sustainable utilization of medicinal plant resources. Attached Figure Description

[0054] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example and not limitation, wherein:

[0055] Figure 1 This is a flowchart illustrating a method for evaluating the growth cycle of medicinal plants based on AI image recognition, provided in an embodiment of the present invention. Detailed Implementation

[0056] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make the invention more thorough and complete, and to fully convey the scope of the invention to those skilled in the art.

[0057] Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0058] It should be noted that the number of any elements in the accompanying drawings is for illustrative purposes only and not as a limitation, and any naming is for distinction only and has no limiting meaning.

[0059] The following is for reference. Figure 1 , Figure 1 This is a flowchart illustrating a method for assessing the growth cycle of medicinal plants based on AI image recognition, provided in an embodiment of the present invention. Figure 1 As shown, a method for assessing the growth cycle of medicinal plants based on AI image recognition includes:

[0060] S1. Acquire multi-dimensional image sequences of the growth process of medicinal plants using multispectral imaging equipment, and simultaneously record ambient light intensity and temperature and humidity data;

[0061] S2. Perform spatiotemporal registration on the multi-dimensional image sequence to eliminate pixel shifts caused by plant swaying and changes in shooting angle;

[0062] S3. An adaptive hybrid filtering algorithm is used to eliminate noise in the registered image while preserving the texture details of the leaf edges.

[0063] S4. Perform multi-scale contrast enhancement on the denoised image and reconstruct the high-frequency feature map through Laplacian pyramid decomposition.

[0064] S5. Construct a dual-branch deep network. The first branch extracts the dynamic feature set of leaf morphology, and the second branch extracts the feature set of physiological parameter changes.

[0065] S6. Perform cross-modal fusion of the morphological dynamic feature set and the physiological parameter change feature set, and calculate the dynamic growth index matrix in combination with environmental parameters.

[0066] S7. Based on the matching degree analysis between the dynamic growth index matrix and the preset growth stage model, output the current growth cycle stage identifier and stage transition probability curve.

[0067] It should be noted that this invention acquires multi-dimensional image sequences of medicinal plant growth processes using a multispectral imaging device, while simultaneously recording ambient light intensity and temperature / humidity data. A multispectral imaging device is capable of simultaneously acquiring image information across multiple spectral bands, capturing the image characteristics of medicinal plants at different wavelengths, thus providing richer information for subsequent analysis. A multi-dimensional image sequence refers to an image sequence acquired at different times, angles, and lighting conditions; these image sequences comprehensively reflect the growth status of the medicinal plant. Ambient light intensity refers to the light intensity in the plant's growth environment, which significantly impacts photosynthesis and growth. Temperature and humidity data refer to the temperature and humidity data in the environment, which also significantly affect plant growth and development. By simultaneously recording these data, the relationship between plant growth and environmental factors can be better analyzed.

[0068] Specifically, multispectral imaging equipment can include various types of cameras, such as visible light cameras, near-infrared cameras, and mid-infrared cameras, each capable of acquiring image information in different wavelength bands. For example, a visible light camera can acquire images of medicinal plants in the visible light band for observing their morphological characteristics; a near-infrared camera can acquire images of plants in the near-infrared band for analyzing their physiological state. When acquiring image sequences, different shooting times and angles can be set according to the growth cycle and characteristics of the medicinal plant. For instance, in the early stages of plant growth, images can be taken at regular intervals to observe germination and growth; in the middle stages of growth, the shooting frequency can be increased to better monitor the plant's growth dynamics. Ambient light intensity can be measured using a light sensor, and temperature and humidity data can be measured using a temperature and humidity sensor. These sensors can work synchronously with the multispectral imaging equipment to ensure data accuracy and consistency. The measurement range of the light sensor can be adjusted according to the actual ambient light intensity. For example, in an indoor planting environment, the light intensity range may be between 100 and 1000 lux; in an outdoor planting environment, the light intensity range may be between 1000 and 100,000 lux. The measurement range of the temperature and humidity sensor is typically -20℃ to 50℃ and 0% to 100% RH. These parameters can be optimized according to the specific planting environment.

[0069] Preferably, to improve the accuracy and reliability of image acquisition, an autofocus function can be introduced into the multispectral imaging device to ensure that a clear image is obtained every time. Simultaneously, the acquired image sequence can be preprocessed, for example, by using denoising algorithms to remove random noise from the images and color correction algorithms to correct color deviations, thereby improving image quality and usability. Furthermore, to better synchronize the recording of ambient light intensity and temperature / humidity data, a timestamp synchronization mechanism can be set between the sensor and the imaging device to ensure temporal consistency between the acquired image data and environmental data.

[0070] In some embodiments, the specific steps for extracting the dynamic morphological feature set of the leaf in step S5 include:

[0071] S51. Perform dynamic deformation-aware convolution operation on the visible light band image sequence. The convolution kernel adaptively adjusts the spatial sampling position according to the leaf growth direction to generate the leaf contour deformation field.

[0072] S52. Based on the deformation field, calculate the three-dimensional curvature integral value and establish the blade curl index as shown in the following formula:

[0073]

[0074] in, This is an index of blade curl. These represent the first and second principal curvatures of the blade surface, respectively; S is the integral region of the blade surface. The area differential;

[0075] S53. Jointly encode the leaf curl index and the leaf vein fractal dimension to generate the morphological dynamic feature set.

[0076] It should be noted that the specific steps for extracting the dynamic morphological feature set of leaves in this invention include performing a dynamic deformation-aware convolution operation on the visible light image sequence. The convolution kernel adaptively adjusts the spatial sampling position according to the leaf growth direction, thereby generating a leaf contour deformation field. The leaf contour deformation field refers to a feature field reflecting the dynamic changes in leaf growth, obtained by calculating the shape changes of the leaf at different time points. Based on this deformation field, a three-dimensional curvature integral value is calculated to establish a leaf curl index, which is used to quantify the degree of leaf curling. The vein fractal dimension is used to describe the complexity and distribution characteristics of the veins. The leaf curl index and the vein fractal dimension are jointly encoded to finally generate a dynamic morphological feature set for subsequent analysis and processing.

[0077] Specifically, dynamic deformation-aware convolution is a special type of convolution operation that dynamically adjusts the sampling position of the convolution kernel according to the leaf's growth direction. In visible light image sequences, the leaf's growth direction can be extracted using image processing algorithms, such as edge detection and morphological analysis. The sampling position of the convolution kernel is adjusted according to the leaf's growth direction, allowing the convolution operation to better capture the leaf's morphological changes. The three-dimensional curvature integral value is obtained by calculating the curvature of the leaf surface, reflecting the degree of leaf bending. During the calculation, the first and second principal curvatures of the leaf surface are key parameters, which can be obtained through geometric analysis of the image. The calculation of the vein fractal dimension involves the complexity analysis of the vein structure, which can be implemented using relevant algorithms from fractal theory. The joint encoding process fuses the leaf curl index and the vein fractal dimension to form a comprehensive dynamic morphological feature set, providing a more complete description of the leaf's morphological characteristics.

[0078] Preferably, when performing dynamic deformation-aware convolutional operations, a more refined leaf growth direction detection algorithm can be introduced, such as combining it with a convolutional neural network (CNN) in deep learning to more accurately identify the leaf growth direction. This can further improve the accuracy of the convolution kernel sampling position, thereby better capturing the morphological changes of the leaf. When calculating the three-dimensional curvature integral value, a multi-scale analysis method can be used to calculate the leaf curvature at different scales to more comprehensively reflect the degree of leaf bending. In addition, for the calculation of the fractal dimension of leaf veins, different fractal analysis methods, such as box counting or related fractal dimension methods, can be considered to adapt to different types of leaf vein structures. In the joint encoding process, more advanced feature fusion techniques, such as attention-based fusion methods, can be used to better highlight important features and improve the quality and effectiveness of the morphological dynamic feature set.

[0079] In some embodiments, the specific steps for extracting the physiological parameter change feature set in step S5 include:

[0080] S54. Construct a dynamic decoupling model of spectral reflectance to separate the chlorophyll absorption band and the overtone vibrational signal of water molecules;

[0081] S55. Based on the spectral fingerprint database of medicinal plant species, calculate the metabolic activity index as shown in the following formula:

[0082]

[0083] in, It is a metabolic activity index; Let be the spectral reflectance function for time t; Serves as a species-specific absorption template; , The characteristic spectral range;

[0084] S56. Perform time-domain cross-correlation analysis on the metabolic activity index and the transpiration fluctuation rate to generate the feature set of physiological parameter changes.

[0085] It should be noted that the specific steps for extracting the physiological parameter change feature set in this invention include constructing a dynamic decoupling model of spectral reflectance and separating the chlorophyll absorption band and the overtone vibrational signals of water molecules. The dynamic decoupling model of spectral reflectance is a model used to analyze changes in plant spectral reflectance. It can decompose complex spectral reflectance signals into different components, thereby extracting features related to the plant's physiological state. The chlorophyll absorption band refers to the specific wavelengths of light absorbed by chlorophyll in the spectrum; changes in reflectance in these bands are closely related to the plant's photosynthetic state. The overtone vibrational signals of water molecules are related to the plant's water state. By separating these signals, the plant's physiological state can be assessed more accurately. Based on the spectral fingerprint database of medicinal plant species, a metabolic activity index is calculated, which is used to quantify the plant's metabolic activity level. Finally, a time-domain cross-correlation analysis is performed on the metabolic activity index and transpiration fluctuation rate to generate a physiological parameter change feature set for subsequent growth cycle assessment.

[0086] Specifically, the construction of the dynamic decoupling model for spectral reflectance requires a large amount of spectral data, which can be obtained from multispectral imaging equipment. The model separates the chlorophyll absorption band and the overtone vibrational signals of water molecules by analyzing the changes in spectral reflectance across different wavelengths. The chlorophyll absorption band is typically located in the red and near-infrared light bands, while the overtone vibrational signals of water molecules appear in specific mid-infrared bands. The calculation of the metabolic activity index requires the use of a spectral fingerprint database of medicinal plant species, which contains the spectral characteristics of different species under different physiological states. By comparing the data with the data in the fingerprint database, the metabolic activity index can be calculated, reflecting the intensity of the plant's metabolic activity over a specific time period. Transpiration volatility refers to the rate of change of plant transpiration over time, which can be calculated by measuring the amount of water evaporated from plant leaves. Temporal cross-correlation analysis is a statistical method used to analyze the correlation between two time series. This method can combine the metabolic activity index with transpiration volatility to generate a set of physiological parameter change characteristics.

[0087] Preferably, to improve the accuracy of the dynamic decoupling model of spectral reflectance, machine learning algorithms, such as Support Vector Machines (SVM) or Convolutional Neural Networks (CNNs) in deep learning, can be introduced to train and optimize the model. These algorithms can learn more accurate decoupling patterns based on a large amount of sample data, thereby more effectively separating the chlorophyll absorption band and the overtone vibrational signals of water molecules. When calculating the metabolic activity index, more spectral feature parameters, such as spectral slope and spectral curvature, can be considered to more comprehensively reflect the metabolic state of plants. In addition, for the measurement of transpiration fluctuation rate, more advanced sensor technologies, such as high-precision humidity and temperature sensors, can be used to improve the accuracy and reliability of the measurement. In the time-domain cross-correlation analysis, more complex statistical models, such as the Autoregressive Moving Average (ARMA) model or wavelet analysis methods, can be used to more accurately capture the dynamic relationship between the metabolic activity index and the transpiration fluctuation rate, thereby generating a higher-quality set of physiological parameter change features.

[0088] In some embodiments, the cross-modal fusion in step S6 employs:

[0089] S61. Construct a growth state coupling model, map the morphological dynamic feature set to the physiological parameter space, and calculate the feature compatibility matrix.

[0090] S62. Generate growth constraint factors based on environmental parameters, wherein the factors include the modulation coefficient of light intensity on photosynthetic phase and the damping coefficient of temperature and humidity on metabolic rate.

[0091] S63. Perform multi-source feature constraint optimization as shown in the following formula:

[0092]

[0093] in, This is the fused dynamic growth index matrix; A set of morphological dynamic features; A set of features representing changes in physiological parameters; , For adaptive balance parameters; For environmental constraint functions; The growth index matrix to be optimized; It is a set of environmentally related factors.

[0094] It should be noted that the cross-modal fusion step in this invention is achieved by constructing a growth state coupling model. This model maps the morphological dynamic feature set to the physiological parameter space, calculates the feature compatibility matrix, and thus evaluates the degree of matching between morphological features and physiological parameters. The growth constraint factors generated by environmental parameters include the modulation coefficient of light intensity on photosynthetic phase and the damping coefficient of temperature and humidity on metabolic rate. These factors are used to reflect the influence of environmental conditions on plant growth. Finally, through multi-source feature constraint optimization, the morphological dynamic feature set, the physiological parameter change feature set, and the environmental constraint factors are comprehensively processed to generate a fused dynamic growth index matrix, which is used to evaluate the growth status of medicinal plants.

[0095] Specifically, the growth status coupling model is a mathematical model used to transform information from the morphological dynamic feature set into the physiological parameter space for comprehensive analysis. The feature compatibility matrix is ​​obtained by calculating the similarity or correlation between morphological features and physiological parameters, reflecting the degree of matching between the two features. The growth constraint factors generated by environmental parameters are calculated based on environmental conditions such as light intensity, temperature, and humidity. The modulation coefficient of light intensity on the photosynthetic phase reflects the influence of light intensity on plant photosynthesis, while the damping coefficients of temperature and humidity on metabolic rate reflect the inhibitory or promoting effects of temperature and humidity on plant metabolic rate. Multi-source feature constraint optimization is a comprehensive process that balances the relationships between the morphological dynamic feature set, the physiological parameter change feature set, and the environmental constraint factors to obtain a fused dynamic growth index matrix, which can more comprehensively reflect the growth status of medicinal plants.

[0096] Preferably, to improve the accuracy of the growth state coupling model, deep learning methods, such as neural networks, can be used to learn the complex mapping relationship between morphological dynamic feature sets and physiological parameters. When calculating the feature compatibility matrix, more feature similarity measurement methods, such as cosine similarity and Euclidean distance, can be introduced to more accurately assess the matching degree between features. For growth constraint factors generated by environmental parameters, more environmental parameters, such as carbon dioxide concentration and soil moisture, can be considered to more comprehensively reflect the impact of the environment on plant growth. In the multi-source feature constraint optimization process, more advanced optimization algorithms, such as genetic algorithms or particle swarm optimization algorithms, can be used to more effectively balance the relationship between different feature sets, thereby obtaining a more accurate dynamic growth index matrix.

[0097] In some embodiments, the environmental constraint function The calculation includes:

[0098] S611. The phase modulation equation for photosynthetically active radiation is established as shown in the following formula:

[0099]

[0100] in, L is the illumination phase modulation factor; L is the real-time illumination intensity. The light saturation intensity of the species; The day-night cycle parameter is t; t is the current time.

[0101] S612. Construct the metabolic damping term as shown in the following formula:

[0102]

[0103] in, The damping factor is the temperature and humidity; T is the ambient temperature. The optimal growth temperature; H represents the temperature tolerance range; H represents the ambient humidity. This represents the critical humidity threshold.

[0104] It should be noted that the calculation of the environmental constraint function in this invention is based on the phase modulation equation of photosynthetically active radiation and the metabolic damping term. The phase modulation equation of photosynthetically active radiation describes the effect of light intensity on the phase of plant photosynthesis, while the metabolic damping term describes the inhibitory effect of ambient temperature and humidity on the plant's metabolic rate. Through these two equations, an environmental constraint factor can be generated, which includes the modulation coefficient of light intensity on the photosynthetic phase and the damping coefficient of temperature and humidity on the metabolic rate. These factors can quantify the impact of environmental conditions on plant growth.

[0105] Specifically, the phase modulation equation for photosynthetically active radiation is a mathematical model that considers factors such as real-time light intensity, species light saturation intensity, diurnal cycle parameters, and the current time to calculate the light phase modulation factor. This factor reflects the regulatory effect of light intensity changes on the photosynthetic phase of plants; for example, at low light intensity, the photosynthetic phase may be delayed. The metabolic damping term is a function related to ambient temperature and humidity, considering factors such as ambient temperature, optimum growth temperature, temperature tolerance range, ambient humidity, and humidity threshold to calculate the temperature and humidity damping factor. This factor reflects the inhibitory or promoting effects of temperature and humidity on plant metabolic rates; for example, at excessively high temperatures or excessively low humidity, the plant's metabolic rate may decrease. The environmental constraint factors calculated using these two equations can provide important environmental parameter support for subsequent growth constraint analysis.

[0106] Preferably, when calculating the phase modulation equation for photosynthetically active radiation, more environmental light parameters, such as the spectral distribution of light, can be incorporated to more accurately reflect the impact of light on plant photosynthesis. For the metabolic damping term, other environmental factors such as soil moisture can be considered to more comprehensively assess the impact of the environment on plant metabolism. Furthermore, machine learning methods can be used to model and optimize the environmental constraint function. By training the model with a large amount of environmental and plant growth data, it can more accurately predict environmental constraint factors. For example, algorithms such as random forests or neural networks can be used to learn the complex relationship between environmental parameters and plant growth based on historical data, thereby improving the accuracy and reliability of the environmental constraint function.

[0107] In some embodiments, the specific implementation of step S7 includes:

[0108] S71. Construct a quantized model of the growth stage, map the dynamic growth index matrix to Hilbert space, and generate discrete energy level basis vectors.

[0109] S72. The coherence between the current dynamic growth index matrix and the basis vectors of each energy level is calculated as shown in the following formula:

[0110]

[0111] in, Let be the coherence of the k-th energy level; This is the growth state density matrix; For the k-th energy level projection operator; Represents the matrix trace operation;

[0112] S73. Generate the stage transition probability curve based on the coherence, and use the Hilbert space basis vectors as topological constraints for the state transition path.

[0113] It should be noted that the generation of the stage transition probability curve in this invention is based on a quantized model of Hilbert space. Hilbert space is a mathematical space used to describe the evolution and interaction of quantum states. In this invention, discrete energy level basis vectors are generated by mapping the dynamic growth index matrix to Hilbert space. These energy level basis vectors are used to represent different stages of plant growth, while coherence is used to measure the similarity between the current growth state and each energy level basis vector. By calculating the coherence, a stage transition probability curve can be generated, which reflects the probability of the plant transitioning from one growth stage to another.

[0114] Specifically, the Hilbert space is constructed based on photosynthetic eigenstates, which reflect the photosynthetic characteristics of plants at different growth stages. The environmental parameter perturbation Hamiltonian is used to introduce the influence of environmental factors on plant growth. By splitting the Hilbert space basis vectors into energy levels, the growth state of plants under different environmental conditions can be reflected more accurately. The imaginary time evolution algorithm is a method for solving the Hamiltonian eigenstates. This algorithm can determine the stage transition path, i.e., the possible path for a plant to transition from one growth stage to another. The coherence is calculated through matrix trace operations, reflecting the similarity between the growth state density matrix and the energy level projection operator. The generation of the stage transition probability curve is based on the coherence calculation results. By analyzing the potential barrier height and thermal fluctuation energy between adjacent energy levels, the calculation of the stage transition probability curve can be further optimized.

[0115] Preferably, to improve the accuracy of the Hilbert space model, more environmental parameters, such as soil moisture and carbon dioxide concentration, can be introduced to more comprehensively reflect the impact of the environment on plant growth. When constructing a quantized model of growth stages, more advanced quantum mechanical methods, such as quantum Monte Carlo methods, can be employed to more accurately solve for the eigenstates of the Hamiltonian. Furthermore, the calculation of stage transition probability curves can be combined with machine learning algorithms, such as Bayesian networks or Markov chain Monte Carlo methods, to more accurately predict the transition probabilities of plant growth stages. For example, by analyzing growth stage transition patterns in historical data, machine learning models can be trained to improve the prediction accuracy of stage transition probability curves.

[0116] In some embodiments, the calculation method of the stage transition probability curve includes:

[0117] S74. Calculate the barrier height between adjacent energy levels based on the Hilbert space basis vectors. Thermal fluctuation energy ;

[0118] S75. The quantum tunneling rate equation is established as shown in the following formula:

[0119]

[0120] in, This refers to the stage transition rate; This refers to the intrinsic transition rate; The height of the barrier; This refers to thermal fluctuation energy. is Boltzmann's constant; T is the ambient temperature.

[0121] It should be noted that the calculation of the stage transition probability curve in this invention is based on Hilbert space basis vectors. Hilbert space basis vectors are a concept in quantized models used to represent different stages of plant growth. By calculating the barrier height and thermal fluctuation energy between adjacent energy levels, a quantum tunneling rate equation can be established. This equation describes the rate at which a plant transitions from one growth stage to another, taking into account the influence of factors such as ambient temperature on the transition process. Through these calculations, the transition probability of plant growth stages can be predicted more accurately.

[0122] Specifically, Hilbert space basis vectors are generated through a quantized model of growth stages, reflecting the characteristic states of plants at different growth stages. The potential barrier height between adjacent energy levels represents the energy barrier required for a plant to transition from one stage to another, while thermal fluctuation energy is related to ambient temperature, reflecting the promoting effect of heat energy on plant growth stage transitions. The quantum tunneling rate equation, based on quantum mechanics, combines factors such as barrier height, thermal fluctuation energy, and ambient temperature to calculate the rate of plant growth stage transitions. The intrinsic transition rate is a constant related to the plant species, representing the rate of plant growth stage transitions under ideal conditions. Through the calculation and analysis of these parameters, stage transition probability curves can be obtained, thus providing an important basis for assessing the plant growth cycle.

[0123] Preferably, when calculating the barrier height between adjacent energy levels, more plant physiological parameters, such as the plant's metabolic activity index, can be incorporated to more comprehensively reflect changes in plant growth status. For the calculation of thermal fluctuation energy, specific environmental conditions, such as humidity and light intensity, can be considered to more accurately assess the impact of environmental factors on plant growth stage transitions. Furthermore, the parameters in the quantum tunneling rate equation can be calibrated and optimized using experimental data to improve the model's accuracy and reliability. For example, growth experiments can be conducted on different species of medicinal plants to collect relevant data on stage transitions, thereby determining the optimal values ​​for parameters such as intrinsic transition rates.

[0124] In some embodiments, the adaptive hybrid filtering algorithm in step S3 includes:

[0125] S31. Perform anisotropic diffusion filtering for blade guidance. The diffusion tensor is shown in the following formula:

[0126]

[0127] Where D is the diffusion tensor; ∇ is the gradient operator; and I is the image gray value. , These are the smoothness intensity parameters for the main vein and secondary veins of the leaf, respectively.

[0128] S32. Perform morphological wavelet packet decomposition in the frequency domain to retain sub-band components that match the vibration frequency of the leaf edge.

[0129] It should be noted that the adaptive hybrid filtering algorithm in step S3 of this invention comprises two main operations: blade-guided anisotropic diffusion filtering and morphological wavelet packet decomposition in the frequency domain. Blade-guided anisotropic diffusion filtering is an image processing technique designed to smooth the image based on the structural features of the blade while preserving the texture details of the blade edges. Morphological wavelet packet decomposition is a frequency domain analysis method used to extract sub-band components that match the vibration frequency of the blade edge, thereby further enhancing the feature information of the image. By combining these two methods, important image details can be preserved while removing noise, providing high-quality input for subsequent image analysis.

[0130] Specifically, leaf-guided anisotropic diffusion filtering is implemented by defining a diffusion tensor that adjusts the smoothing intensity based on the directions of the midrib and secondary veins of the leaf. The smoothing intensity parameter determines the degree of smoothing in different directions; for example, the smoothing intensity along the midrib may be higher to reduce noise, while the smoothing intensity perpendicular to the midrib remains lower to preserve edge details. Morphological wavelet packet decomposition is performed in the frequency domain, extracting sub-band components within different frequency ranges by performing multi-level decomposition of the image. Leaf edge vibration frequencies refer to the minute vibration frequencies that may occur at the leaf edge during natural growth. By preserving sub-band components matching these frequencies, the feature information of the leaf edge in the image can be enhanced, making it more suitable for subsequent morphological analysis.

[0131] Preferably, when implementing leaf-guided anisotropic diffusion filtering, an adaptive method can be used to dynamically adjust the smoothing intensity parameter. For example, the smoothing intensity can be automatically adjusted based on the local structural features of the leaves in the image to better adapt to the noise levels and texture features of different regions. In morphological wavelet packet decomposition, more decomposition layers can be introduced to extract feature information in different frequency ranges more precisely. Furthermore, machine learning methods, such as convolutional neural networks (CNNs), can be combined to extract and analyze features from the decomposed sub-band components, further improving the quality of image features. For example, a CNN model can be trained to identify and enhance features related to leaf growth status, thereby providing more accurate image data for subsequent growth cycle assessment.

[0132] In some embodiments, the dynamic deformation-aware convolution operation includes:

[0133] S511. The deformation offset of the convolution kernel is generated based on the leaf growth direction field, as shown in the following formula:

[0134]

[0135] in, This is the kernel offset; This refers to the deformation sensitivity parameter; For the leaf vein gradient field; For random disturbance terms;

[0136] S512. Apply the offset to the convolution kernel coordinates through differentiable sampling to generate the deformed feature map.

[0137] It should be noted that the dynamic deformation-aware convolution operation in this invention generates a convolution kernel deformation offset based on the leaf growth direction field, and applies this offset to the convolution kernel coordinates to generate a deformed feature map. Dynamic deformation-aware convolution is an advanced image processing technique designed to dynamically adjust the shape and position of the convolution kernel according to the leaf's growth direction, thereby more accurately capturing changes in leaf morphology. The convolution kernel deformation offset is calculated based on the leaf growth direction field, reflecting the differences in the leaf's growth direction at different positions. By applying the offset to the convolution kernel coordinates through differentiable sampling, a feature map consistent with the leaf's growth direction can be generated, thus better reflecting the dynamic changes of the leaf.

[0138] Specifically, the leaf growth direction field refers to the distribution of the leaf's growth direction at different locations, which can be extracted from multispectral images using image processing algorithms. The calculation of the convolution kernel deformation offset involves the deformation sensitivity parameter and the vein gradient field. The deformation sensitivity parameter is an adjustment factor used to control the degree of convolution kernel deformation; the vein gradient field reflects the distribution and direction of veins within the leaf and is a crucial basis for calculating the offset. By applying the offset to the convolution kernel coordinates, dynamic deformation of the convolution kernel can be achieved, allowing it to better adapt to the leaf's growth direction. Differentiable sampling is a numerical calculation method that allows dynamic adjustment of the sampling position during convolution operations, thereby generating a deformed feature map. This feature map can more accurately reflect the morphological changes of the leaf, providing higher-quality input for subsequent morphological feature extraction.

[0139] Preferably, when calculating the convolution kernel deformation offset, more image features, such as leaf texture and color features, can be incorporated to more comprehensively reflect the leaf's growth state. For example, combining texture features can better capture subtle changes on the leaf surface, while color features help distinguish different types of leaf tissue. When implementing differentiable sampling, more advanced numerical calculation methods, such as Gaussian sampling or adaptive sampling, can be employed to improve sampling accuracy and efficiency. Furthermore, deep learning techniques, such as convolutional neural networks (CNNs), can be combined to optimize dynamic deformation-aware convolution operations. By training a CNN model, the optimal convolution kernel deformation strategy can be automatically learned, thereby further improving the accuracy and robustness of feature extraction.

[0140] In some embodiments, the growth stage quantization model includes:

[0141] S711. Construct a growth Hilbert space whose basis vectors are composed of photosynthetic eigenstates;

[0142] S712, Introducing the environmental parameter perturbation Hamiltonian The Hilbert space basis vectors are split into energy levels.

[0143] S713. Solve the Hamiltonian using the imaginary time evolution algorithm. The eigenstates of the given state determine the stage transition path as shown in the following formula:

[0144]

[0145] in, Let be the quantum state at time t; is the initial state; e is the natural exponential function.

[0146] It should be noted that the quantized model of the growth stages in this invention is constructed by introducing a perturbation Hamiltonian based on environmental parameters. The Hamiltonian is a fundamental concept in quantum mechanics, used to describe the energy state of a system. In this invention, the perturbation Hamiltonian is used to account for the influence of environmental factors on the plant growth stages. By solving for the eigenstates of the Hamiltonian using the imaginary time evolution algorithm, the transition paths of the plant growth stages can be determined. This method introduces quantum mechanical concepts into the assessment of plant growth stages, providing a new theoretical foundation for the dynamic assessment of the plant growth cycle.

[0147] Specifically, the quantized growth stage model is constructed based on the Hilbert space of plant growth, where the basis vectors of the Hilbert space are composed of photosynthetic eigenstates. These eigenstates reflect the photosynthetic characteristics of plants at different growth stages. The perturbation Hamiltonian of environmental parameters considers the influence of environmental factors such as light intensity, temperature, and humidity on plant growth. By introducing the perturbation Hamiltonian, the basis vectors of the Hilbert space can undergo energy level splitting, thus more accurately reflecting the plant's growth state under different environmental conditions. The imaginary time evolution algorithm is a numerical method for solving the Hamiltonian eigenstates, which can determine the transition path of plant growth stages. This method can map the plant's growth state to the evolution process of quantum states, thus providing a new perspective for the dynamic evaluation of plant growth stages.

[0148] Preferably, when constructing a quantum model of the growth stage, more environmental parameters, such as soil moisture and carbon dioxide concentration, can be introduced to more comprehensively reflect the impact of the environment on plant growth. When introducing the perturbation Hamiltonian of the environmental parameter, more complex quantum mechanical models, such as many-body interaction models, can be considered to more accurately describe the physicochemical changes during plant growth.

[0149] Furthermore, the parameter settings of the imaginary time evolution algorithm can be optimized based on specific plant species and growth environments. For example, calibrating the time step and evolution time in the algorithm using experimental data can improve the accuracy and reliability of the model. Machine learning techniques, such as Generative Adversarial Networks (GANs) in deep learning, can also be combined to optimize and validate the quantized growth stage model. These methods can further improve the accuracy and robustness of plant growth stage assessment.

[0150] The above-described embodiments of the present invention have the following beneficial effects: The present invention utilizes a multispectral imaging device to acquire multidimensional image sequences of the growth process of medicinal plants, and simultaneously records ambient light intensity and temperature and humidity data, comprehensively acquiring morphological and physiological information of plant growth, providing rich data for growth cycle assessment. Spatiotemporal registration is used to eliminate pixel offset, improving the accuracy and consistency of image data; an adaptive hybrid filtering algorithm is used to remove noise, retain leaf edge texture details, and improve image quality; multi-scale contrast enhancement is used to highlight key features and enhance feature recognizability. A dual-branch deep network is constructed to extract dynamic morphological feature sets and physiological parameter change feature sets of leaves respectively, reflecting the plant growth state in multiple dimensions; the two feature sets are fused across modalities and combined with environmental parameters to calculate a dynamic growth index matrix, comprehensively considering morphological, physiological characteristics, and environmental influences, making the growth index more representative and accurate; based on the matching degree analysis between the dynamic growth index matrix and the preset growth stage model, the current growth cycle stage identifier and stage transition probability curve are output, realizing real-time monitoring and prediction of the growth stage of medicinal plants, providing support for scientific management and precise regulation.

[0151] When extracting the dynamic morphological feature set of leaves, this invention generates a leaf contour deformation field through dynamic deformation-sensing convolution operations, calculates the leaf curl index, and co-encodes this index with the fractal dimension of leaf veins to precisely capture leaf morphological changes and enrich morphological feature information. When extracting the feature set of physiological parameter changes, a dynamic decoupling model of spectral reflectance is constructed to separate the chlorophyll absorption band and the overtone vibration signal of water molecules, calculates the metabolic activity index, and performs time-domain cross-correlation analysis with the transpiration volatility to accurately reflect changes in plant physiological state. In the cross-modal fusion process, a growth state coupling model is constructed, a feature compatibility matrix is ​​calculated, growth constraint factors are generated, and multi-source feature constraint optimization is performed to better integrate morphological and physiological features, while also considering environmental constraints, making the dynamic growth index matrix more scientific and practical. When calculating the environmental constraint function, a phase modulation equation for photosynthetically active radiation and a metabolic damping term are established to accurately quantify the impact of environmental factors on plant growth. In the calculation of stage transition probability curves, a quantized model of growth stages is constructed, coherence is calculated, and stage transition probability curves are generated. Hilbert space basis vectors are used as topological constraints for state transition paths, accurately predicting plant growth stage transitions based on quantum mechanics principles. In the adaptive hybrid filtering algorithm, leaf-guided anisotropic diffusion filtering and morphological wavelet packet decomposition in the frequency domain are performed to effectively remove noise and preserve leaf edge texture details. In the dynamic deformation-aware convolution operation, a convolution kernel deformation offset is generated based on the leaf growth direction field. A deformed feature map is generated through differentiable sampling, flexibly capturing leaf morphological changes. In the quantized model of growth stages, a growth Hilbert space is constructed, and environmental parameter perturbations of the Hamiltonian are introduced. The eigenstates of the Hamiltonian are solved using a virtual-time evolution algorithm to determine the stage transition path. Based on the framework of quantum mechanics, the mechanism of plant growth stage transitions is studied in depth, providing a solid theoretical foundation for accurate assessment of the growth cycle.

[0152] Furthermore, the storage medium in the embodiments of this application stores program instructions capable of implementing all the above methods. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0153] The above description is merely an explanation of some preferred embodiments of the present invention and the technical principles employed. Those skilled in the art should understand that the scope of the invention as described in the embodiments of the present invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A method for assessing the growth cycle of medicinal plants based on AI image recognition, characterized in that, Includes the following steps: S1. Acquire multi-dimensional image sequences of the medicinal plant growth process using a multispectral imaging device, and simultaneously record ambient light intensity and temperature and humidity data; S2. Perform spatiotemporal registration on the multi-dimensional image sequences to eliminate pixel shifts caused by plant swaying and changes in shooting angle; S3. An adaptive hybrid filtering algorithm is used to eliminate noise in the registered image while preserving leaf edge texture details. S4. Multi-scale contrast enhancement is performed on the denoised image, and high-frequency feature maps are reconstructed through Laplacian pyramid decomposition. S5. A dual-branch deep network is constructed, with the first branch extracting the leaf morphological dynamic feature set and the second branch extracting the physiological parameter change feature set. S6. The morphological dynamic feature set and the physiological parameter change feature set are fused across modes, and a dynamic growth index matrix is ​​calculated in conjunction with environmental parameters. S7. Based on the matching degree analysis between the dynamic growth index matrix and the preset growth stage model, the current growth cycle stage identifier and stage transition probability curve are output.

2. The method according to claim 1, characterized in that, The specific steps for extracting the dynamic morphological feature set of the leaf in step S5 include: S51, performing a dynamic deformation-aware convolution operation on the visible light band image sequence, wherein the convolution operation adaptively adjusts the spatial sampling position according to the leaf growth direction to generate a leaf contour deformation field; S52, calculating the three-dimensional curvature integral value based on the deformation field to establish the leaf curl degree; S53, jointly encoding the leaf curl degree and the leaf vein fractal dimension to generate the dynamic morphological feature set.

3. The method according to claim 1, characterized in that, The specific steps for extracting the physiological parameter change feature set in step S5 include: S54, constructing a dynamic decoupling model of spectral reflectance to separate the chlorophyll absorption band and the overtone vibration signal of water molecules; S55, calculating the metabolic activity index based on the spectral fingerprint database of medicinal plant species; S56, performing time-domain cross-correlation analysis on the metabolic activity index and the transpiration fluctuation rate to generate the physiological parameter change feature set.

4. The method according to claim 1, characterized in that, The cross-modal fusion in step S6 adopts the following steps: S61, constructing a growth state coupling model, mapping the morphological dynamic feature set to the physiological parameter space, and calculating the feature compatibility matrix; S62, generating growth constraint factors based on environmental parameters, wherein the factors include the modulation coefficient of light intensity on photosynthetic phase and the damping coefficient of temperature and humidity on metabolic rate. S63. Perform multi-source feature constraint optimization as shown in the following formula: ,in, This is the fused dynamic growth index matrix; For morphological dynamic feature set; A set of features representing changes in physiological parameters; , For adaptive balance parameters; For environmental constraint functions; The growth index matrix to be optimized; It is a set of environmentally related factors.

5. The method according to claim 4, characterized in that, The environmental constraint function The calculation includes: S611, establishing the phase modulation equation for photosynthetically active radiation as shown in the following formula: ,in, L is the illumination phase modulation factor; L is the real-time illumination intensity. The light saturation intensity of the species; S612. The metabolic damping term is constructed as shown in the following formula: ,in, The damping factor is the temperature and humidity; T is the ambient temperature. The optimal growth temperature; H represents the temperature tolerance range; H represents the ambient humidity. This is the critical threshold for humidity.

6. The method according to claim 1, characterized in that, The specific implementation of step S7 includes: S71, constructing a quantized model of the growth stage, mapping the dynamic growth index matrix to Hilbert space, and generating discrete energy level basis vectors; S72, calculating the coherence between the current dynamic growth index matrix and each energy level basis vector; S73, generating a stage transition probability curve based on the coherence, and using the Hilbert space basis vectors as topological constraints for the state transition path.

7. The method according to claim 6, characterized in that, The calculation method for the stage transition probability curve includes: S74, calculating the potential barrier height between adjacent energy levels based on the Hilbert space basis vectors. Thermal fluctuation energy S75. The quantum tunneling rate equation is established as shown in the following formula: ,in, This refers to the stage transition rate; This refers to the intrinsic transition rate; The height of the barrier; This refers to thermal fluctuation energy; is Boltzmann's constant; T is the ambient temperature.

8. The method according to claim 1, characterized in that, The adaptive hybrid filtering algorithm in step S3 includes: S31, performing anisotropic diffusion filtering for blade guidance; S32. Perform morphological wavelet packet decomposition in the frequency domain to retain sub-band components that match the vibration frequency of the leaf edge.

9. The method according to claim 2, characterized in that, The dynamic deformation-aware convolution operation includes: S511, generating a convolution kernel deformation offset based on the leaf growth direction field; S512. Apply the offset to the convolution kernel coordinates through differentiable sampling to generate the deformed feature map.

10. The method according to claim 6, characterized in that, The quantum model of the growth stage includes: S711, constructing a growth Hilbert space whose basis vectors are composed of photosynthetic eigenstates; S712, introducing environmental parameter perturbation Hamiltonian to split the Hilbert space basis vectors into energy levels; S713, solving for the eigenstates of the Hamiltonian through imaginary time evolution algorithm to determine the stage transition path.

Images