Leaf Disease and Pest Identification Method and System Based on Image Processing

Abstract

The present invention relates to the field of agricultural science and technology, specifically a method and system for identifying leaf diseases and pests based on image processing, including the following steps: setting up multispectral and thermal imaging devices to collect multispectral images and thermal imaging data in the target frequency band, deleting images with insufficient clarity and data with excessive noise to obtain optimized multispectral images, and marking and classifying the data according to the collection date to form a time series image dataset. In the present invention, by dynamically adjusting network parameters to reflect the states of each stage of leaf growth, the prediction of diseases and pests is not only based on static data, but reflects the actual changes in plant growth, improving the real-time application value and accuracy of the prediction model. The introduction of state transfer learning enables the transition from the healthy stage to the stage of diseases and pests to be predicted more accurately, for early diagnosis and timely intervention, improving the recognition accuracy of the specific location of diseases and pests, enabling local treatment, and effectively enhancing the pertinence and economic efficiency of treatment.

Description

Technical Field

[0001] The present invention relates to the field of agricultural science and technology, and in particular to a leaf pest identification method and system based on image processing. Background Art

[0002] The field of agritech focuses on using modern scientific and technological means to improve and enhance agricultural production processes. It covers a wide range of fields, from precision agriculture and remote sensing monitoring to genetic engineering and plant pathology. In agritech, special attention is paid to developing technologies that can improve crop yield, quality and stress resistance while reducing environmental impact and costs. Technological advances in this field include the application of robotics in agriculture, the development of smart agricultural systems, and the optimization and protection of crops through information technology and biotechnology.

[0003] Among them, the leaf pest and disease recognition method refers to the use of image processing technology to automatically detect and identify pests and diseases on crop leaves. The main purpose is to enable farmers to take timely measures to prevent and control pests and diseases through early identification of pests and diseases, protecting crops from major losses. Through this technology, the use of chemical pesticides can be reduced, the health level and overall yield of crops can be improved, and it also helps protect the environment.

[0004] In practical applications, existing technologies rely on traditional image processing technology and primary data analysis, and fail to fully utilize time series data and advanced model algorithms. This leads to deficiencies in real-time monitoring and accurate prediction of pest and disease development, especially in the identification and prediction of early pest and disease stages. Due to the lack of high-contrast and dynamically adjusted data processing mechanisms, existing technologies cannot accurately capture subtle changes, which is crucial in plant pathology because early intervention is the key to preventing the spread of pests and diseases. The analysis of state transitions in existing technologies is not sophisticated enough to provide an accurate path for pest and disease progression, limiting the timeliness and effectiveness of decision-making. Technical limitations not only increase crop losses, but also lead to over-reliance on chemical pesticides, increasing the burden on the environment. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a leaf disease and insect pest identification method based on image processing.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a leaf pest identification method based on image processing, comprising the following steps:

[0007] S1: Set up multispectral and thermal imaging equipment, collect multispectral images and thermal imaging data of the target frequency band, delete images with insufficient clarity and data with too much noise, obtain optimized multispectral images, and mark and classify the data according to the collection date to form a time series image dataset;

[0008] S2: Use the time series image dataset to record the feature changes at the target moment, dynamically adjust the Bayesian network according to the feature changes, record the adjusted weight and bias parameters, match the growth status of the leaves in the differentiation stage, and construct an adjusted dynamic model;

[0009] S3: Perform state transition learning on the adjusted dynamic model, calculate the transition probability between stages through the transition probability between states, and obtain a state transition probability matrix;

[0010] S4: Use the state transition probability matrix to extract the pest and disease areas on the leaves in the image, identify the development trend of pests and diseases, predict the development trend of pests and diseases in the future time period, obtain the probability value of the differential prediction path, and evaluate the influence range of the development of pests and diseases through probability distribution, and output the prediction result of the development of pests and diseases;

[0011] S5: Based on the prediction result of the development of pests and diseases, input the affected area features into the conditional random field model, identify and mark the positions of pests and diseases on the leaves, analyze the spatial dependence of image pixels, and output the identification and location of pests and diseases.

[0012] As a further solution of the present invention, the step of obtaining the optimized multispectral image is specifically as follows:

[0013] S111: Set up multispectral and thermal imaging devices, adjust the sensing range to a preset target frequency band, perform device performance detection, and analyze the stability of signal transmission to obtain an initialized multispectral image and thermal imaging data;

[0014] S112: Use image denoising technology to process the initialized multispectral image and thermal imaging data, and apply image enhancement technology to optimize the image edges to obtain processed multispectral image and thermal imaging data;

[0015] S113: According to the processed multispectral image and thermal imaging data, use the formula:

[0016] ;

[0017] Calculate the weighted sum and corrected quality index of each data , and obtain an optimized multispectral image;

[0018] Among them, represents the intensity of the th data point, represents the weight of the data point, represents the distance from the data point to the target center, represents the attenuation constant, represents the correction coefficient, and e is the natural constant.

[0019] As a further solution of the present invention, the steps for obtaining the time series image dataset are specifically as follows:

[0020] S121: Record the time stamp for each image of the optimized multi-spectral image, associate the time information with the image data, and sort the multi-spectral image data with time stamps in chronological order to obtain a multi-band image dataset;

[0021] S122: Based on the multi-band image dataset, use the formula:

[0022] ;

[0023] Calculate the weighted time series value to obtain the time series image dataset;

[0024] Wherein, represents the weighted time series value, represents the time stamp of the th image, is the weight assigned according to the image quality score, represents the minimum time stamp of the image, represents the maximum time stamp of the image, is the number of images.

[0025] As a further solution of the present invention, the steps for obtaining the adjusted weight and bias parameters are specifically as follows:

[0026] S211: Based on the time series image dataset, record the feature changes at the target moment, evaluate the probability distribution of the impact on plant growth, and generate an initial probability evaluation result;

[0027] S212: Use the initial probability evaluation result to update the weight and bias parameters through the Bayesian decision theory to obtain the updated Bayesian network parameters;

[0028] S213: Through the updated Bayesian network parameters, use the formula:

[0029] ;

[0030] Calculate the current weight parameter to obtain the adjusted weight and bias parameters;

[0031] Wherein, represents the old weight parameter, is the learning rate, is the error between the target and the prediction, is the current feature change index, is the regularization coefficient, is the attenuation factor.

[0032] As a further solution of the present invention, the obtaining steps of the state transition probability matrix are specifically as follows:

[0033] S311: Perform state transition learning on the adjusted dynamic model, analyze the state set of the adjusted dynamic model, including healthy, initial infection, and severe infection, identify the basic structure of the model, and generate a state definition set;

[0034] S312: Based on the differentiated states in the state definition set, divide the time periods when the differentiated states appear, integrate the occurrence frequencies of each differentiated state, identify the transition probabilities between the differentiated states through the Markov chain algorithm, and obtain the initial transfer probability data;

[0035] S313: According to the initial transfer probability data, perform data collation, count the transfer frequencies of the differentiated states, and use the formula:

[0036] ;

[0037] Obtain the transition probability matrix;

[0038] wherein, represents the probability of transitioning from state to state , represents the number of observations of transitioning from state to state , is the total number of observations of transitioning from state to the remaining states, is the correction factor, is the environmental factor, is the stability coefficient of state , is the observation window size.

[0039] As a further solution of the present invention, the obtaining steps of the probability value of the differentiated prediction path are specifically as follows:

[0040] S411: According to the adjusted state transition probability matrix, extract the pest and disease areas on the leaves in the image and perform grayscale processing, identify the pest and disease diffusion process of the leaves, and obtain the pest and disease development sequence;

[0041] S412: Using the pest and disease development sequence, apply the Markov chain theory to each pest and disease diffusion process, calculate the path probabilities of reaching multiple end states starting from the current state, and generate a probability distribution map;

[0042] S413: Based on the probability distribution diagram, analyze the probability distribution of each path, select the associated paths with deviations, and use the formula:

[0043] ;

[0044] Obtain the probability value of the differential prediction path ;

[0045] where, is the improved state transition probability of pests and diseases, is the weight coefficient, is the variability factor of state transition under the target environment, is the time since the last pest and disease event, is the decay constant, is the number of states, is the natural constant.

[0046] As a further solution of the present invention, the steps for obtaining the prediction result of the development of pests and diseases are specifically as follows:

[0047] S421: Use the probability value of the differential prediction path, identify the pest and disease area through edge detection and region segmentation, identify the development trend of pests and diseases, and identify the pest and disease paths with serious development trends to obtain the key pest and disease path set;

[0048] S422: Apply a deep learning model to perform time series analysis on the key pest and disease path set, combine environmental factors and real-time pest and disease development data, predict the development trend of pests and diseases for each path, and obtain pest and disease evolution trend data;

[0049] S423: Verify the pest and disease evolution trend data, adjust the parameters of the prediction model by comparing with the pest and disease development data at the fitting time point, optimize the consistency of the prediction, and use the formula:

[0050] ;

[0051] Obtain the prediction result of the development of pests and diseases;

[0052] where, is the prediction result of the development of pests and diseases, is the weight coefficient, and represent the prediction and real-time data, and are the non-linear adjustment coefficients, is the weight of the influence of the current environmental factor on the prediction, is the current environmental condition index.

[0053] As a further solution of the present invention, the acquisition steps of the pest and disease identification and positioning are specifically as follows:

[0054] S511: Based on the pest and disease development prediction result, input the characteristics of the affected area, including vegetation index and spectral data, into the conditional random field model to generate a regional feature model;

[0055] S512: Use the regional feature model for scanning, identify and mark the positions of pests and diseases on the leaves, and generate a leaf image identifying the pests and diseases;

[0056] S513: Adopt the leaf image identifying the pests and diseases, Calculate the weighted dependence impact evaluation value based on the dependence between each pixel and its neighborhood pixels, sort and classify the pixels in the leaf pest and disease image using the evaluation value, identify the spatial position of the pest and disease area, record the pest and disease distribution area, and use the formula:

[0057] ;

[0058] Output the pest and disease identification and positioning;

[0059] Wherein, represents the weighted dependence impact evaluation value, represents the total number of pixels in the image, represents the th pixel weight, represents the th pixel's dependence on its surrounding pixels, is the adjustment coefficient for pixel proximity.

[0060] A leaf pest and disease identification system based on image processing, which is used to execute the above-mentioned leaf pest and disease identification method based on image processing. The system includes:

[0061] The image acquisition module, based on a multispectral device and a thermal imaging device, acquires multispectral images and thermal imaging data, deletes invalid data with insufficient image clarity and excessive noise, obtains an optimized multispectral image, classifies and marks the data, and constructs a time series image dataset;

[0062] The feature change dynamic modeling module, based on the time series image dataset, extracts the feature change information at the target moment, dynamically adjusts the Bayesian network, updates the weight and bias parameters according to the feature change, generates an adjusted dynamic model, and records the weight, bias, and correlation parameters during the state transition process through the model to obtain the state transition learning result;

[0063] The state transition module uses the state transition learning result to calculate the state transition probability for different growth stages of the leaves, obtains the state transition probability matrix, extracts the pest and disease areas of the leaves, identifies the development trend of pests and diseases, and outputs the prediction result of the development of pests and diseases.

[0064] Based on the prediction result of the development of pests and diseases, the pest and disease location tracking module inputs the affected area features into the conditional random field model to identify and calibrate the locations of pests and diseases, and outputs the identification and location of pests and diseases.

[0065] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0066] In the present invention, the accuracy of pest and disease identification and the reliability of prediction are optimized through time series image data and feature change records. After collecting multi-spectral images and thermal imaging data, high-contrast images are screened and labeled for classification, enhancing the quality of the data and the effectiveness of subsequent processing, and providing a solid foundation for establishing an accurate model. By dynamically adjusting the network parameters to reflect the states of each stage of leaf growth, the pest and disease prediction is not only based on static data but reflects the actual changes in plant growth, improving the real-time application value and accuracy of the prediction model. The introduction of state transition learning enables the transition from the healthy stage to various pest and disease stages to be predicted more accurately, which is crucial for early diagnosis and timely intervention, reducing crop losses and dependence on chemical pesticides. The application of the conditional random field model based on the prediction result further improves the identification accuracy of the specific locations of pests and diseases, enabling local treatment and effectively enhancing the pertinence and economic efficiency of the treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0068] Figure 2 It is a flow chart of the steps for obtaining the optimized multi-spectral image of the present invention;

[0069] Figure 3 It is a flow chart of the steps for obtaining the time series image data set of the present invention;

[0070] Figure 4 It is a flow chart of the steps for obtaining the adjusted weight and bias parameters of the present invention;

[0071] Figure 5 It is a flow chart of the steps for obtaining the state transition probability matrix of the present invention;

[0072] Figure 6 It is a flow chart of the steps for obtaining the probability value of the differential prediction path of the present invention;

[0073] Figure 7 It is a flow chart of the steps for obtaining the prediction result of the development of pests and diseases of the present invention;

[0074] Figure 8 This is the flowchart of the acquisition steps for pest and disease identification and location in the present invention. Specific embodiments

[0075] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0076] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more unless otherwise specifically defined.

[0077] Please refer to Figure 1 , the present invention provides a technical solution: a method for identifying pests and diseases on leaves based on image processing, including the following steps:

[0078] S1: Set up multi-spectral and thermal imaging devices, collect multi-spectral images and thermal imaging data in the target frequency band, delete images with insufficient clarity and data with excessive noise, obtain optimized multi-spectral images, and mark and classify the data according to the collection date to form a time-series image dataset;

[0079] S2: Use the time-series image dataset to record the characteristic changes at the target moment, dynamically adjust the Bayesian network according to the characteristic changes, record the adjusted weight and bias parameters, match the growth conditions of the leaves in the differential stage, and construct an adjusted dynamic model;

[0080] S3: Perform state transition learning on the adjusted dynamic model, calculate the transition probability between stages through the transition probability between states, and obtain the state transition probability matrix;

[0081] S4: Use the state transition probability matrix to extract the pest and disease areas on the leaves in the image, identify the development trend of pests and diseases, predict the development trend of pests and diseases in the future time period, obtain the probability value of the differential prediction path, and evaluate the influence range of the development of pests and diseases through the probability distribution, and output the prediction result of the development of pests and diseases;

[0082] S5: Based on the prediction results of the development of pests and diseases, input the characteristics of the affected areas into the conditional random field model, identify and mark the positions of pests and diseases on the leaves, analyze the spatial dependence of image pixels, check the consistency of the positioning, and output the identification and positioning of pests and diseases;

[0083] The optimized multi-spectral images include enhanced color discrimination, texture resolution, and target feature recognition; the time-series image dataset includes multi-spectral images, thermal imaging data, and images with high contrast; the adjusted weight and bias parameters include the weight adjustment value, bias adjustment value, and the influence of the target output feature on the output; the adjusted dynamic model includes the updated Bayesian network weights, updated Bayesian network bias parameters, and simulation of the leaf growth status; the state transition probability matrix includes the transition probability from healthy to initial infection, the transition probability from initial infection to severe infection, and the transition probability of maintaining the current state; the probability values of the differential prediction paths include the probability of the development stage of pests and diseases, the probability distribution map, and the spatial range of the spread of pests and diseases; the prediction results of the development of pests and diseases include the characteristics of the affected areas, the marked positions of pests and diseases, and the analysis results of pixel-level spatial dependence, and the identification and positioning of pests and diseases include the types of pests and diseases identified on the leaves, the area of the leaves affected by pests and diseases, and the position coordinates of the marked pests and diseases on the leaves.

[0084] Please refer to Figure 2 , and the specific steps for obtaining the optimized multi-spectral images are as follows:

[0085] S111: Set up multi-spectral and thermal imaging devices, adjust the sensing range to the preset target frequency band, conduct device performance detection, and analyze the stability of signal transmission to obtain the initialized multi-spectral images and thermal imaging data;

[0086] The process of setting up multi-spectral and thermal imaging devices involves adjusting the devices to conform to the preset target frequency band and performing basic settings of the devices, including selecting appropriate spectral band ranges and temperature ranges. According to the device manual or actual needs, set the appropriate band and temperature parameters for the multi-spectral and thermal imaging instruments, and adjust through the device's debugging interface to ensure that the sensor response meets expectations. It is necessary to verify the working status by detecting the performance of the sensor. Comparative experiments can be used, such as comparing with a known standard source or calibrating with a target object, to ensure that the image quality captured by the device and the thermal imaging data meet the requirements. During performance detection, focus on the sensitivity of the sensor, image clarity, and the accuracy of the thermal image, collect signals and analyze the stability of signal transmission in real time, and check whether there are large fluctuations or interferences in the signals. Through multiple tests, ensure stable signal transmission and obtain the initialized multi-spectral images and thermal imaging data.

[0087] S112: Process the initialized multi-spectral image and thermal imaging data using image denoising techniques, and apply image enhancement techniques to optimize the image edges to obtain the processed multi-spectral image and thermal imaging data;

[0088] After obtaining the initialized multi-spectral image and thermal imaging data, data processing is carried out to improve the image quality. Removing noise from the data is an essential step. By using digital filtering techniques, random noise in the data is screened out, which involves performing spectral analysis on the data, identifying the frequency components unique to the noise and suppressing them. Applying image enhancement techniques to optimize the image edges includes sharpening and contrast enhancement. Sharpening makes the image look clearer by increasing the contrast in areas where the color changes rapidly in the image, and contrast enhancement optimizes the dynamic range of the entire image by adjusting the dark and bright parts of the image to obtain the processed multi-spectral image and thermal imaging data.

[0089] S113: According to the processed multi-spectral image and thermal imaging data, use the formula:

[0090]

[0091] Calculate the weighted sum and corrected quality index of each data , to obtain the optimized multi-spectral image;

[0092] Among them, represents the intensity of the th data point, represents the weight of the data point, represents the distance from the data point to the target center, represents the attenuation constant, represents the correction coefficient, and e is the natural constant,

[0093] The specific meanings of the parameters in the formula are as follows: is the intensity of the th data point, expressed as the amplitude of the data point signal;

[0094] is the weight of the data point, considering the signal-to-noise ratio and importance of the data;

[0095] is the actual distance from the data point to the target center, used to measure the spatial contribution of the data point;

[0096] is an attenuation constant, representing the rate at which the data weakens with increasing distance;

[0097] is a correction coefficient that adjusts the influence of each data point according to environmental factors. To calculate the specific value of the parameter, it needs to be set according to on-site measurements and equipment calibration data;

[0098] For example, it can be set by measuring the response curve and noise level of the device, then it can be calculated based on the average line-of-sight distance and light attenuation properties of the site, it needs to be adjusted according to actual environmental conditions such as temperature, humidity, and air pressure. The specific value is set as unit, , meter, meter, , and substituting it into the formula for calculation gives:

[0099] ;

[0100] This result indicates that the quality index after weighted sum correction is higher than the preset threshold, confirming the high quality of the data.

[0101] Please refer to Figure 3 , the steps for obtaining the time series image dataset are specifically as follows:

[0102] S121: Record the timestamp for each image of the optimized multispectral image, associate the time information with the image data, and sort the multispectral image data with timestamps in chronological order to obtain a multi-band image dataset;

[0103] Recording the timestamp for the optimized multispectral image is mainly to ensure that the collection time of each image is accurately marked. The key to this process lies in the setting of the image data management system. It is necessary to confirm whether the time synchronization function of the system is enabled to ensure that the system time is synchronized with the Coordinated Universal Time (UTC). During the image collection process, the system will automatically attach the current time as a timestamp to the metadata of each image file. This step is automatically completed through programming and helps to track and monitor any anomalies that occur during the data collection process.

[0104] Using the multispectral image data with timestamps, data sorting and verification is a systematic data management activity. By reading the timestamp information in the metadata of each image, the images are sorted in chronological order. This automated process ensures that all images are arranged according to the actual collection timeline. Checking the continuity and time consistency of the sorted data includes comparing whether the intervals between image timestamps conform to the predetermined collection frequency and whether there are anomalies such as time jumps or overlaps. This verification process is a key step to ensure data integrity and availability, and a multi-band image dataset is obtained.

[0105] S122: Based on the multi - band image dataset, use the formula:

[0106] ;

[0107] Calculate the weighted time - series value to obtain the time - series image dataset;

[0108] Among them, represents the weighted time - series value, represents the timestamp of the th image, is the weight assigned according to the image quality score, represents the minimum timestamp of the image, represents the maximum timestamp of the image, is the number of images.

[0109] An improved time - weight assignment formula is used. This formula combines the timestamp and the image quality score to calculate the weighted time value of each image, so as to better reflect the importance of each image in the time - series dataset. represents the weighted time - series value, is the weight assigned based on the image quality score, providing the relative importance of the image in the dataset. , and are the minimum and maximum timestamps in the dataset respectively, used to normalize the timestamp so that the calculated weights are distributed between 0 and 1. Suppose there are three images with timestamps minutes, and the corresponding image quality scores result in weight assignments of , and the minimum timestamp minutes, the maximum timestamp minutes. The normalized timestamp is calculated as:

[0110] For : ;

[0111] For : ;

[0112] For : ;

[0113] The weighted time value is calculated as:

[0114] ;

[0115] This result shows that after weighting and normalization, a weighted time - series value that comprehensively considers time and image quality is obtained, providing a basis for subsequent time - series analysis.

[0116] Please refer to Figure 4 , the steps for obtaining the adjusted weight and bias parameters are specifically as follows:

[0117] S211: Based on the time series image dataset, record the feature changes at the target moment, evaluate the probability distribution of the impact on plant growth, and generate an initial probability evaluation result;

[0118] The selection of the target moment is based on a predetermined growth cycle or significant environmental changes. This process involves complex data screening and feature recognition techniques, using computer vision and machine learning algorithms to automatically identify key changes in the images. For example, edge detection techniques are used to identify the emergence of new leaves or flowers in the plant growth state. By setting thresholds, the key stages of plant growth, such as flowering or fruit ripening, can be identified. Each labeled image will be further analyzed to evaluate the impact of the changes on plant growth, estimate the occurrence probabilities of different growth stages, and generate an initial probability evaluation result.

[0119] S212: Use the initial probability evaluation result to update the weight and bias parameters through Bayesian decision theory to obtain the updated Bayesian network parameters;

[0120] Using Bayesian decision theory to update the weight and bias parameters, the process involves using the initial probability evaluation result, applying Bayes' theorem to calculate the posterior probability. On this basis, the weight and bias parameters in the network are adjusted to better reflect the new data evidence. The process includes the calculation of prior probability and likelihood probability, as well as how to update the network parameters through probability. Through an iterative process, the prediction accuracy of the entire network is optimized to obtain the updated Bayesian network parameters.

[0121] S213: Through the updated Bayesian network parameters, use the formula:

[0122] ;

[0123] Calculate the current weight parameter , and obtain the adjusted weight and bias parameters;

[0124] Among them, represents the old weight parameter, is the learning rate, is the error between the target and the prediction, is the current feature change index, is the regularization coefficient, is the decay factor;

[0125] Adjust the weight parameter of the Bayesian network. In the formula represents the new weight parameter, is the previous weight parameter, represents the learning rate, represents the error between the target and the prediction, represents the current feature change index, is the regularization coefficient, is the weight decay factor;

[0126] Set to obtain the current feature change index from the dataset , the error between the target and the prediction , the learning rate , the regularization coefficient , the weight decay factor , and the initial weight , then the process of calculating the new weight is as follows:

[0127] ;

[0128] This result indicates that by responding to the current feature change and considering weight decay, the new weight parameter has increased slightly, indicating that the model has gradually adapted to the feature change, which helps to more accurately predict the future change trend.

[0129] Please refer to Figure 5 , the steps for obtaining the state transition probability matrix are specifically as follows:

[0130] S311: Conduct state transition learning on the adjusted dynamic model, analyze the state set of the adjusted dynamic model, including healthy, initial infection, and severe infection, identify the basic structure of the model, and generate a state definition set;

[0131] State transition learning involves analyzing the state set of the adjusted model. During the process, it is necessary to identify and define various health states, such as healthy, initial infection, and severe infection. By using state monitoring and diagnostic techniques, such as physiological signal monitoring and biomarker detection, to accurately identify the states, analyze the characteristic data of the states to establish clear state definitions, and involve using machine learning methods, such as clustering analysis or decision trees, to distinguish the key features of different states for supporting the dynamic tracking and updating of states, including the management of data streams, and generate a state definition set.

[0132] S312: Based on the differentiated states in the state definition set, divide the time periods when the differentiated states occur, integrate the occurrence frequencies of each differentiated state, and identify the transition probabilities between the differentiated states through the Markov chain algorithm to obtain the initial transition probability data;

[0133] To understand the transition rules between states, it is necessary to collect sufficient real-time data. The data reflects the conversion situations between different states. Statistical methods or machine learning algorithms, such as Markov models, are used to analyze the data to identify the probabilities of transitioning from one state to another. The process involves the estimation of probability distributions and parameter optimization to ensure the accuracy and reliability of the model, providing quantifiable dynamic transition logic, which is the key for the model to predict future state changes, and obtaining the initial transition probability data.

[0134] S313: According to the initial transition probability data, conduct data collation, count the transition frequencies of different states, and use the formula:

[0135] ;

[0136] Obtain the transition probability matrix;

[0137] Among them, represents the probability of transitioning from state to state , represents the number of observations of transitioning from state to state , is the total number of observations of transitioning from state to the remaining states, is the correction factor, is the environmental factor, reflecting the external influence on the target state , is the stability coefficient of state , is the observation window size;

[0138] Set the number of observations (healthy) to state (initial infection) within a specific observation period to be 100 times, and the total number of times of transitioning from the healthy state to any state to be 300 times. Set the correction factor to 1.2, the environmental factor to 10, the stability coefficient of state to 0.9, and the observation window size to 20. Then the formula calculation is as follows:

[0139] ;

[0140] ​This result shows that under the given parameters and conditions, the transition probability from the healthy state to the initial infection state is approximately 44.83%. This result indicates that after adjusting for the influence of the observed data, the probability of transitioning from the healthy state to the initial infection state is significant, which helps in making more accurate risk assessments and formulating preventive measures in practical applications.

[0141] Please refer to Figure 6 , and the specific steps for obtaining the probability values of the differential prediction paths are as follows:

[0142] S411: According to the adjusted state transition probability matrix, extract the pest and disease areas on the leaves in the image and perform grayscale processing to identify the pest and disease diffusion process on the leaves, obtaining the pest and disease development sequence;

[0143] To extract the pest and disease areas on the leaves in the image, image segmentation techniques need to be applied to separate the leaf area in the input multispectral image from the background. This process can be achieved through threshold-based methods, color space segmentation (such as HSV, LAB), or deep learning models (such as convolutional neural networks) to identify and extract the leaf area. The extracted leaf area contains different color information and needs to be grayscale processed. Using the weighted average method, the RGB image is converted into a grayscale image to ensure that the contribution of different colors to the grayscale is adjusted according to human eye perception. After grayscale processing, the pest and disease information in the image is manifested as different grayscale values or texture changes. Through methods such as edge detection, morphological processing, and region growing, the significantly changed lesion areas in the image can be detected. The pest and disease diffusion process describes the specific diffusion process and pattern of pests and diseases on the leaves, involving image processing and analysis of the affected leaves. For example, through image grayscale processing, the specific route of pests and diseases spreading from one area to another can be identified and traced. By analyzing the characteristics of the lesion areas (such as size, shape, location), the development process of pests and diseases on the leaves can be tracked. Combining with the state transition probability matrix, the transition of pests and diseases from one stage to another can be automatically traced, which includes identifying the initial manifestations, progression speed, and diffusion path of pests and diseases, obtaining the pest and disease development sequence.

[0144] S412: Using the pest and disease development sequence, apply the Markov chain theory to each pest and disease diffusion process, calculate the path probabilities from the current state to multiple end states, and generate a probability distribution map;

[0145] Statistical software is needed to process and analyze the sequence data of the development of pests and diseases. By establishing a Markov model, the transition probabilities of each state are used to predict the development trend of pests and diseases, which refers to the change and development pattern of pests and diseases in a future period of time, including the growth rate of pests and diseases, the expansion of the affected range, and the change in severity. After processing the input data of the model, the calculation process involves setting different end states, such as the alleviation or aggravation of pests and diseases, and estimating the probability of reaching the state, showing the diffusion process and probability of various pests and diseases, providing a scientific basis for formulating control measures, and generating a probability distribution map.

[0146] S413: Based on the probability distribution map, analyze the probability distribution of each path, select the associated paths with deviations, and use the formula:

[0147] ;

[0148] Obtain the probability value of the differential prediction path ;

[0149] Among them, is the improved state transition probability of pests and diseases, is the weight coefficient, is the variability factor of the state transition under the target environment, is the time since the last pest and disease event, is the decay constant, is the number of states, is the natural constant;

[0150] : The transition probability from one state to another;

[0151] : The weight coefficient, considering the influence degree of the state transition;

[0152] : The environmental factor, reflecting the change of state transition under different environmental conditions;

[0153] : The time decay factor, where represents the time interval from the last state transition to the present;

[0154] is the decay constant, reflecting the influence of time;

[0155] In the denominator, is the normalization factor, ensuring that the sum of probabilities is 1;

[0156] There are 3 state transitions set, and the parameters of each are set as follows:

[0157] ;

[0158] ;

[0159] ;

[0160] Calculate terms: ;

[0161] Calculate the numerator: ;

[0162] Calculate the denominator: ;

[0163] Probability calculation:

[0164] ;

[0165] This result shows that the probability value of the differential prediction path that combines environmental factors and time decay is 0.123, which reflects the probability of a specific path occurring under given conditions. This value further guides the formulation of prevention and control strategies for the spread process of high-risk pests and diseases.

[0166] Please refer to Figure 7 , and the specific steps for obtaining the prediction results of the development of pests and diseases are as follows:

[0167] S421: Adopt the probability value of the differential prediction path, identify the pest and disease areas through edge detection and region segmentation, identify the development trend of pests and diseases, identify the pest and disease paths with serious development trends, and obtain the set of key pest and disease paths;

[0168] Identify the edge areas in the image through the edge detection algorithm. Use the Canny edge detection algorithm, which extracts edges by calculating the gradient value of the image and setting thresholds, and further enhances the detailed information in the image through binary processing to complete the preliminary edge recognition. Through region segmentation technology, divide the image into different regions, and segment based on the color difference, texture features, and edge information of adjacent regions. Common region segmentation algorithms include the k-means clustering algorithm and the threshold segmentation algorithm. The obtained regions are the pest and disease areas. Analyze the development trend of pests and diseases for each path. Based on the differential prediction paths in the real-time data and combined with the pest and disease areas obtained after region segmentation, further calculate the pest and disease probabilities of each path at different time points. The identification of pest and disease paths is achieved by matching with the existing pest and disease development models, and identify the paths with the most serious pest and disease development in different regions to obtain the set of key pest and disease paths.

[0169] S422: Apply a deep learning model to perform time series analysis on the set of key pest and disease paths, combine environmental factors and real-time pest and disease development data, predict the development trends of pests and diseases for each path, and obtain pest and disease evolution trend data;

[0170] Perform time series analysis on the set of key pest and disease paths, use a deep learning model, combine environmental factors and real-time pest and disease development data, which involves using a deep neural network to train the model through real-time data to identify pest and disease patterns, taking environmental factors and real-time data as inputs, predicting the development trends of pests and diseases, and the prediction process includes data processing such as normalization and outlier handling, as well as parameter optimization and validation of the model to obtain pest and disease evolution trend data.

[0171] S423: Validate the pest and disease evolution trend data, adjust the parameters of the prediction model by comparing with the pest and disease development data at the fitting time points, optimize the prediction consistency, using the formula:

[0172] ;

[0173] Obtain the pest and disease development prediction result;

[0174] Among them, is the pest and disease development prediction result, is the weight coefficient, and represent the predicted and real-time data, and are non-linear , is the weight of the influence of the current environmental factor on the prediction,

[0175] is the current environmental condition index;

[0176] The specific explanations and setting bases of the parameters in the formula are as follows:

[0177] is the weight coefficient, used to balance the influence of the predicted data and the real-time data , and this weight reflects the confidence level in the prediction model compared to the real-time data. In this model, the weight is determined based on the performance of the previous model and expert evaluation. A higher value indicates that the data of the prediction model is considered relatively reliable;

[0178] and are non-linear adjustment coefficients, used to enhance the influence of the prediction and real-time data respectively. Here , , which means the impact of the predicted data is squared, increasing its weight in the overall assessment, while the real-time data retains its original influence. This setting is optimized according to the non-linear characteristics of the pest and disease development pattern;

[0179] is the current environmental factor 's weight coefficient, which reflects the degree of influence of environmental variables in the overall prediction. Indicates that in the overall calculation, environmental factors are also taken into account. This value is obtained through the analysis of the environmental impact of past data;

[0180] = 0.75 and = 0.5 are obtained through data monitoring and ensemble prediction techniques, reflecting the comparison between the prediction of the future pest and disease development trend and the past actual records;

[0181] = 0.65 is the value of the current environmental conditions, obtained through a real-time monitoring system, including the comprehensive impact index of environmental parameters such as temperature and humidity;

[0182] Substitute the specific values into the formula for calculation:

[0183] ;

[0184] This calculation result indicates that after integrating the predicted data, real-time data, and current environmental factors, the pest and disease development trend index is above medium, which provides a comprehensive perspective to evaluate the future pest and disease development potential and can be used for further formulating prevention and control strategies. This value is closer to the predicted value , indicating that the current environmental conditions and the output of the prediction model are mutually verified to a certain extent.

[0185] Please refer to Figure 8 , the acquisition steps for pest and disease identification and location are specifically as follows:

[0186] S511: Based on the pest and disease development prediction results, input the characteristics of the affected area into the conditional random field model, including the vegetation index and spectral data, to generate a regional feature model;

[0187] Input the characteristics of the affected area into the conditional random field model, and use geographic information system (GIS) and remote sensing technology to obtain accurate data of the affected area, including the leaf damage area, infection type, and infection degree. GIS analyzes multi-temporal satellite images to identify the changed areas, calculates the area and change rate of the areas, and the data is converted into a format acceptable to the conditional random field model. The model analyzes the characteristics to predict the future pest and disease development trend and generates a regional feature model.

[0188] S512: Use the regional feature model to perform scanning, identify and mark the positions of pests and diseases on the leaves, and generate a leaf image identifying pests and diseases;

[0189] Perform high-precision scanning, identify and accurately mark the positions of pests and diseases on the leaves. This process includes using high-resolution image scanning technology and image processing algorithms. The scanning technology can capture the minute details on the leaf surface, and the image processing algorithms analyze the color, texture, and morphology of the image to automatically identify the positions of the infected areas. The algorithm analyzes based on preset pest and disease characteristic parameters, such as typical pest and disease signs like color spots and holes on the leaves. Each mark records the specific position and area, generating a leaf image identifying pests and diseases.

[0190] S513: Adopt the leaf image identifying pests and diseases, Calculate the weighted dependency impact evaluation value based on the dependency between each pixel and its neighboring pixels, use the evaluation value to sort and classify the pixels in the leaf pest and disease image, identify the spatial positions of the pest and disease areas, record the pest and disease distribution areas, using the formula:

[0191] ;

[0192] Output pest and disease identification and positioning;

[0193] Among them, represents the weighted dependency impact evaluation value, represents the total number of pixels in the image, represents the weight of the th pixel, represents the dependency of the th pixel on its surrounding pixels, and

[0194] is the adjustment coefficient for pixel proximity. In this formula, represents the total number of pixels in the image, represents the weight of the th pixel, represents the dependency of the th pixel on its surrounding pixels, is the adjustment coefficient for this pixel's proximity. Assuming an image has 1000 pixels, the weight of each pixel is set to 1, the dependency

[0195] ;

[0196] The result shows that the pixel dependence in the image and the evaluation value of the adjacent adjustment effect are 17.89. This value reflects the consistency and reliability of the image processing algorithm when marking the positions of pests and diseases. The higher the value, the better the consistency of the algorithm, indicating that the calculation and analysis provide a scientific basis for the identification and location of pests and diseases.

[0197] A leaf pest and disease identification system based on image processing. The leaf pest and disease identification system based on image processing is used to execute the above-mentioned leaf pest and disease identification method based on image processing. The system includes:

[0198] An image acquisition module, based on a multispectral device and a thermal imaging device, acquires multispectral images and thermal imaging data, deletes invalid data with insufficient image clarity and excessive noise, obtains an optimized multispectral image, classifies and marks the data, and constructs a time-series image dataset;

[0199] A feature change dynamic modeling module, based on the time-series image dataset, extracts the feature change information at the target moment, dynamically adjusts the Bayesian network, updates the weight and bias parameters according to the feature change, generates an adjusted dynamic model, and records the weight, bias, and correlation parameters in the state transition process through the model to obtain a state transition learning result;

[0200] A state transition module uses the state transition learning result to calculate the state transition probability of different growth stages of the leaf, obtains a state transition probability matrix, extracts the leaf pest and disease area, identifies the development trend of pests and diseases, and outputs a pest and disease development prediction result;

[0201] A pest and disease position tracking module, based on the pest and disease development prediction result, inputs the affected area features into a conditional random field model to identify and calibrate the pest and disease positions, and outputs the pest and disease identification and location.

[0202] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for identifying leaf diseases and pests based on image processing, characterized in that, It includes the following steps: Set up multispectral and thermal imaging devices to collect multispectral images and thermal imaging data in the target frequency band, delete images with insufficient clarity and data with excessive noise to obtain optimized multispectral images, and mark and classify the data according to the collection date to form a time-series image dataset; Use the time-series image dataset to record the characteristic changes at the target moment, dynamically adjust the Bayesian network according to the characteristic changes, record the adjusted weight and bias parameters, match the growth conditions of the leaves in the differentiation stage, and construct an adjusted dynamic model; Perform state transition learning on the adjusted dynamic model, calculate the transition probability between stages through the transition probability between states to obtain a state transition probability matrix; Use the state transition probability matrix to extract the pest and disease areas on the leaves in the image, identify the development trend of pests and diseases, predict the development trend of pests and diseases in the future time period to obtain the probability value of the differential prediction path, and evaluate the influence range of the development of pests and diseases through probability distribution, and output the prediction result of the development of pests and diseases; Based on the prediction result of the development of pests and diseases, input the affected area features into the conditional random field model, identify and mark the positions of pests and diseases on the leaves, analyze the spatial dependence of image pixels, and output the identification and location of pests and diseases.

2. The method for identifying leaf pests and diseases based on image processing according to claim 1, wherein The specific steps for obtaining the optimized multispectral image are as follows: Set up multispectral and thermal imaging devices, adjust the sensing range to the preset target frequency band, perform device performance detection, and analyze the stability of signal transmission to obtain the initialized multispectral image and thermal imaging data; Use image denoising technology to process the initialized multispectral image and thermal imaging data, and apply image enhancement technology to optimize the image edge to obtain the processed multispectral image and thermal imaging data; According to the processed multispectral image and thermal imaging data, use the formula: ； Calculate the weighted sum of each data and the corrected quality index , and obtain the optimized multi-spectral image; Among them, represents the intensity of the th data point, represents the weight of the th data point, represents the distance from the th data point to the target center, represents the attenuation constant of the th data point, represents the correction coefficient, where e is the natural constant.

3. The method for identifying leaf diseases and pests based on image processing according to claim 2, wherein, The specific steps for obtaining the time-series image dataset are as follows: Record the time stamp for each image of the optimized multispectral image, associate the time information with the image data, and sort the multispectral image data with time stamps in chronological order to obtain a multi-band image dataset; Based on the multi-band image dataset, use the formula: ； Calculate the weighted time-series value to obtain a time-series image dataset; in, represents the weighted time series value, Representative The timestamp of the image, is the weight assigned according to the image quality score, Represents the minimum timestamp of the image. Represents the maximum timestamp of the image, is the number of images.

4. The leaf pest and disease identification method based on image processing according to claim 3, wherein, The specific steps for obtaining the adjusted weight and bias parameters are as follows: Based on the time-series image dataset, record the characteristic changes at the target moment, evaluate the probability distribution of the impact on plant growth, and generate an initialized probability evaluation result; Use the initialized probability evaluation result to update the weight and bias parameters through Bayesian decision theory to obtain the updated Bayesian network parameters; Through the updated Bayesian network parameters, use the formula: ； Calculate the current weight parameters , and obtain the adjusted weight and bias parameters; Among them, represents the old weight parameter, is the learning rate, is the error between the target and the prediction, is the current feature change index, is the regularization coefficient, is the decay factor.

5. The method for identifying leaf diseases and pests based on image processing according to claim 4, wherein, The specific steps for obtaining the state transition probability matrix are as follows: Perform state transition learning on the adjusted dynamic model, analyze the state set of the adjusted dynamic model, including healthy, initial infection, and severe infection, identify the basic structure of the model, and generate a state definition set; Based on the differentiated states in the state definition set, divide the time periods when the differentiated states occur, integrate the occurrence frequencies of each differentiated state, and identify the transition probabilities between the differentiated states through the Markov chain algorithm to obtain the initial transition probability data; According to the initial transition probability data, conduct data collation, count the transition frequencies of the differentiated states, and use the formula: ； Obtain the transition probability matrix; Among them, represents the probability of transitioning from state to state . represents the number of observations of transitioning from state to state . is the total number of observations of transitioning from state to the remaining states, is the correction factor, is the environmental factor, is the stability coefficient of state , is the observation window size.

6. The method for identifying leaf diseases and pests based on image processing according to claim 5, characterized in that The steps for obtaining the probability value of the differentiated prediction path are specifically as follows: According to the adjusted state transition probability matrix, extract the pest and disease areas on the leaves in the image and perform grayscale processing, mark the pest and disease diffusion process of the leaves, and obtain the pest and disease development sequence; Using the pest and disease development sequence, apply the Markov chain theory to each pest and disease diffusion process, calculate the path probabilities from the current state to multiple end states, and generate a probability distribution map; Based on the probability distribution map, analyze the probability distribution of each path, select the associated paths with deviations, and use the formula: ； Obtain the probability value of the differentiated prediction path ; Among them, is the improved state transition probability of pests and diseases, is the weight coefficient, is the variability factor of state transition under the target environment, is the time since the last pest and disease event, is the decay constant, is the number of states, is the natural constant.

7. The method for identifying leaf diseases and pests based on image processing according to claim 6, wherein, The steps for obtaining the pest and disease development prediction result are specifically as follows: Adopt the probability value of the differentiated prediction path, identify the pest and disease areas through edge detection and region segmentation, identify the pest and disease development trend, mark the pest and disease paths with serious development trends, and obtain the key pest and disease path set; Apply a deep learning model to conduct time series analysis on the key pest and disease path set, combine environmental factors and real-time pest and disease development data, predict the pest and disease development trends of each path, and obtain the pest and disease evolution trend data; Verify the pest and disease evolution trend data, adjust the parameters of the prediction model by comparing with the pest and disease development data at the fitting time point, and optimize the prediction consistency, using the formula: ； Obtain the pest and disease development prediction result; Among them, is the prediction result of the development of pests and diseases, is the weight coefficient, and represent the predicted and real-time data, and are the non-linear adjustment coefficients, is the weight of the influence of the current environmental factors on the prediction, is the current environmental condition index.

8. The method for identifying leaf diseases and pests based on image processing according to claim 7, characterized in that, The steps for obtaining the pest and disease identification and location are specifically as follows: Based on the pest and disease development prediction result, input the characteristics of the affected area into the conditional random field model, including vegetation index and spectral data, to generate a regional feature model; Use the regional feature model to scan, identify and mark the positions of pests and diseases on the leaves, and generate a leaf image identifying pests and diseases; Adopt the leaf image identifying pests and diseases, analyze the dependence between each pixel and its neighboring pixels, calculate the weighted dependence impact evaluation value, use the evaluation value to sort and classify the pixels in the leaf pest and disease image, identify the spatial positions of the pest and disease areas, and record the pest and disease distribution areas, using the formula: ； Output the pest and disease identification and location; in, represents the weighted dependency impact assessment value, represents the total number of pixels in the image, Indicates The weight of the pixels, Representative The dependence of a pixel on the surrounding pixels, is the adjustment factor of the pixel neighborhood.

9. A leaf pest and disease identification system based on image processing, characterized in that, According to the leaf pest and disease identification method based on image processing according to any one of claims 1-8, the system includes: The image acquisition module, based on a multispectral device and a thermal imaging device, acquires multispectral images and thermal imaging data, deletes invalid data with insufficient image clarity and excessive noise, obtains an optimized multispectral image, and classifies and marks the data to construct a time series image data set; The feature change dynamic modeling module extracts the feature change information at the target moment based on the time series image dataset, dynamically adjusts the Bayesian network, updates the weight and bias parameters according to the feature changes, generates the adjusted dynamic model, and records the weights, biases, and correlation parameters in the state transition process through the model to obtain the state transition learning result; The state transition module uses the state transition learning result to calculate the state transition probability for the differential growth stages of the leaves, obtains the state transition probability matrix, extracts the pest and disease areas of the leaves, identifies the development trend of pests and diseases, and outputs the pest and disease development prediction result; The pest and disease location tracking module inputs the affected area features into the conditional random field model based on the pest and disease development prediction result to identify and calibrate the pest and disease locations, and outputs the pest and disease identification and location.

Images