Meta-learning-based wheat disease and pest intelligent identification method and intelligent identification system

Abstract

The application belongs to the technical field of crop disease and pest identification, and discloses a wheat disease and pest intelligent identification method and an intelligent identification system based on meta learning. The collected leaf and stem images of the wheat to be identified are subjected to filtering processing by using an adaptive Wiener filtering method; the filtered leaf and stem images of the wheat are subjected to gray scale processing to obtain the gray scale images of the leaf and stem of the wheat; the filtered gray scale images of the leaf and stem of the wheat are subjected to enhancement processing by using a cross line gray scale image definition improvement model; and the diseases and pests, symptoms, disease spots and pest body characteristics of the leaf and stem of the wheat are identified according to the enhanced images. The application can be used in agricultural actual production, can provide a reference for further chemical control and biological control of wheat diseases and pests, and can provide a reference for the application of automatic distinguishing and identifying plant stress.

Description

Technical Field

[0001] This invention belongs to the field of crop disease and pest identification technology, and particularly relates to a method and system for intelligent identification of wheat diseases and pests based on meta-learning. Background Technology

[0002] Identifying wheat diseases and pests has always been a challenge in agricultural production. There are many types of wheat diseases and pests, and the treatment measures for different diseases and pests vary, making disease and pest identification an urgent problem for farmers to solve in the production process.

[0003] Traditional pest and disease image recognition methods are mostly limited to the identification of specific pest and disease types, lacking the ability to distinguish between different pests and diseases; existing methods have low recognition efficiency and low recognition accuracy, which limits their practical application.

[0004] Based on the above analysis, the problems and shortcomings of the existing technology are as follows: traditional disease image recognition methods lack the ability to distinguish between different diseases and pests; existing methods have low recognition efficiency and low recognition accuracy; and their practical application is limited. Summary of the Invention

[0005] To overcome the problems existing in related technologies, the embodiments disclosed in this invention provide a method and system for intelligent identification of wheat diseases and pests based on meta-learning.

[0006] The technical solution is as follows: a meta-learning-based intelligent identification method for wheat diseases and pests, including the following steps:

[0007] S1, The adaptive Wiener filtering method is used to filter the collected images of the leaves and stems of the wheat to be identified;

[0008] S2, perform grayscale processing on the filtered wheat leaf and stem images to obtain grayscale images of wheat leaves and stems;

[0009] S3 uses a crosshair grayscale image sharpness improvement model to enhance the grayscale images of wheat leaves and stems after filtering.

[0010] S4. Based on the enhanced image, identify diseases and pests on wheat leaves and stems, symptoms, lesions, and insect characteristics of pests.

[0011] S5. Based on the obtained climate data, soil environmental data, and historical data of the wheat planting area to be identified, a preliminary analysis of wheat diseases and pests is conducted.

[0012] In step S3, the crosshair grayscale image sharpness improvement model is as follows:

[0013]

[0014]

[0015] Where C represents the sharpness of the filtered image; B(I,j) represents the pixel grayscale matrix of the camera at the placement point; m×n represents the number of pixels of the camera at the placement point, 0≤I≤m-1, 0≤J≤n-1, B max B represents the maximum grayscale value of the crosshair grayscale image; min B represents the minimum grayscale value. dif This represents the difference in grayscale values.

[0016] In step S4, the pest and disease identification model for identifying diseases and pests, symptoms, lesions, and insect characteristics of wheat leaves and stems includes: an input layer and a feature extraction network.

[0017] The input layer is used to take the extracted wheat feature image as input to the pest and disease identification model;

[0018] The feature extraction network is used to extract information from the input wheat feature image using a residual network that integrates a multi-layer attention mechanism; and the information extracted by the feature extraction network is normalized using a softmax function.

[0019] In one embodiment, the identification method of the pest and disease identification model includes:

[0020] Step 1: Using multiple complementary features, based on different extraction methods, and heterogeneously formed, images of wheat leaves and stems with diseases, pests, symptoms, lesions, and insect body features are depicted from different visual perspectives. The heterogeneity based on different extraction methods includes: extracting a set of excellent features from multiple perspectives such as shape, texture, and deep learning based on the complementarity between heterogeneous features.

[0021] Step 2 involves in-depth mining of similar pest and disease characteristics embedded among different features to achieve real-time feature fusion. This in-depth mining of similar pest and disease characteristics embedded among different features includes: mining similar pest and disease characteristics through DCA model, CCL model, DCCA model, SCCA model, MLP model, CCA model, feature splicing, and multi-kernel learning fusion model. Among them, similar pest and disease characteristics include: different features all pointing to the same or similar pests and diseases, which are represented by SG, SV, SR, GH, GV, GR, and SD respectively, for a total of 7 groups of similar pest and disease characteristics.

[0022] Step 3 involves dynamically calculating the pest and disease occurrence probability weights of the features, completing feature fusion and optimizing feature combinations to achieve post-feature fusion, and finally realizing deep fusion for multi-faceted pest and disease occurrence probability prediction. The specific method for completing feature fusion and optimizing feature combinations to achieve post-feature fusion includes: using a hybrid soft and hard voting mechanism to vote on the feature combinations generated by the fusion, performing SUM fusion, and simultaneously making decisions based on Max. Two strategies are designed: random or sorted. The top n groups of pest and disease occurrence probability feature combinations with the best accuracy are selected from the n′ groups of feature combinations, where n = 11, n′ = 3, 5, 7, 9, 11. The prediction results are then subjected to soft and hard voting. Two voting decisions are used to integrate and learn different feature discrimination results to achieve feature fusion in the later stage. The specific implementation method of the fusion includes: calculating the estimated probability of single feature and similar pest and disease features based on the Adaboost algorithm; improving the effective region gene selection algorithm to dynamically calculate the pest and disease occurrence probability weights of single feature and similar pest and disease features, using the pest and disease occurrence probability weights to estimate the probability, constructing feature combinations, and using the improved pest and disease occurrence probability model to achieve multi-feature fusion; in the soft and hard voting decisions on the prediction results, the decision may adopt a stacking boosting fusion algorithm or a weighted fusion strategy.

[0023] Step 4 involves identifying images of diseases and pests, symptoms, lesions, and insect characteristics on wheat leaves and stems, and outputting predicted labels based on the identification results.

[0024] In one embodiment, the preliminary analysis of wheat diseases and pests specifically includes the following steps:

[0025] First, historical information on wheat diseases and pests was obtained, and climate and soil environmental data of the wheat planting areas to be identified were analyzed.

[0026] Secondly, based on the analysis results of climate data and soil environment data of the wheat planting area, combined with the suitable environment for wheat diseases and pests, the diseases and pests that may occur in the wheat in the current area are predicted.

[0027] Finally, based on the historical information on wheat diseases and pests and the prediction of wheat diseases and pests that may occur in the current region, preliminary analysis results on wheat diseases and pests were obtained.

[0028] In one embodiment, the preliminary analysis results of wheat diseases and pests specifically include:

[0029] (1) The first wheat disease and pest is obtained by taking the intersection of the historical wheat disease and pest information and the predicted wheat diseases and pests that will occur in the current area.

[0030] (2) Assign weights to the historical wheat pest and disease information and the predicted wheat pest and disease in the current area;

[0031] (3) Determine the weights of various diseases and pests included in the historical information on wheat diseases and pests and the data on diseases and pests that will occur in wheat in the current region.

[0032] (4) Output the data of the first wheat disease and pest, as well as other diseases and pests arranged in order of weight, as the analysis results.

[0033] Another objective of this invention is to provide a meta-learning-based intelligent identification system for wheat diseases and pests, used to implement the aforementioned meta-learning-based intelligent identification method for wheat diseases and pests, wherein the meta-learning-based intelligent identification system for wheat diseases and pests includes:

[0034] The central control module uses a microcontroller or controller to control the normal operation of each module.

[0035] The image preprocessing module, connected to the central control module, is used to preprocess the acquired images of the leaves and stems of the wheat to be identified.

[0036] The image feature extraction module, connected to the central control module, is used to extract features from the preprocessed image;

[0037] The pest and disease identification model building module is connected to the central control module and is used to build pest and disease identification models.

[0038] The meta-learning model building module, connected to the central control module, is used to build the meta-learning model for pest and disease identification.

[0039] The preliminary analysis module for diseases and pests is connected to the central control module and is used to conduct preliminary analysis of wheat diseases and pests based on the acquired climate data, soil environmental data and historical data of the wheat planting area to be identified.

[0040] The soil testing module, connected to the central control module, is used to collect and test soil data from wheat roots at sampling points.

[0041] The pest and disease identification module is connected to the central control module and is used to identify wheat pests and diseases based on extracted wheat image features using an optimized pest and disease model.

[0042] The identification result output module is connected to the central control module and is used to output intelligent identification results of wheat diseases and pests based on the preliminary identification results of wheat diseases and pests, soil test results, and identification of disease and pest models.

[0043] In one embodiment, the meta-learning-based intelligent identification system for wheat diseases and pests further includes:

[0044] The environmental data acquisition module, connected to the central control module, is used to collect climate data, soil environmental data, and overall planning data for the wheat planting area to be identified.

[0045] The historical data acquisition module, connected to the central control module, is used to obtain historical pest and disease information and data of wheat in the current planting area from a professional database.

[0046] The sampling point and camera point setting module is connected to the central control module and is used to set the sampling points of soil samples and the placement points of camera equipment based on the overall planning data of the planting area to be identified.

[0047] The image acquisition module, connected to the central control module, is used to acquire images of the leaves and stems of the wheat to be identified using the camera equipment at the placement point;

[0048] The model training module, connected to the central control module, is used to train the constructed pest and disease model and the meta-learning model of the pest and disease model using the acquired historical data.

[0049] The optimization module, connected to the central control module, is used to optimize the parameters of the pest and disease identification model using the trained meta-learning model.

[0050] Combining all the above technical solutions, the advantages and positive effects of this invention are as follows:

[0051] First, addressing the technical problems and difficulties in solving the aforementioned existing technologies, and closely combining the technical solution to be protected by this invention with the results and data from the research and development process, this invention provides a detailed and in-depth analysis of how the technical solution of this invention solves the technical problems and the creative technical effects brought about after solving the problems. Specifically, this invention utilizes a residual network incorporating a multi-layer attention mechanism for wheat pest and disease identification, which shortens the identification time while improving the identification accuracy and the model's generalization performance, enabling the identification of various wheat pest and disease images. Simultaneously, this invention applies meta-learning methods to the optimization of the pest and disease identification model, avoiding model overfitting, improving the training speed of the pest and disease identification model, and reducing the server computing power cost required for model training, optimization, and updates.

[0052] Secondly, considering the technical solution as a whole or from the perspective of the product, the technical effects and advantages of the technical solution to be protected by this invention are specifically described as follows: This invention can distinguish different wheat diseases and pests based on their characteristics, and has good versatility; at the same time, it has high identification efficiency and high accuracy, and can be used in actual agricultural production, providing a reference for further chemical and biological control of wheat diseases and pests, and can also provide a reference for the application of automatic identification of plant stress.

[0053] Third, this invention utilizes multiple complementary features in the pest and disease identification model, based on different extraction methods, to heterogeneously depict the symptoms, lesions, and insect body features of wheat leaves and stems from different visual perspectives; it deeply mines the similar pest and disease features contained between different features to achieve real-time feature fusion; it dynamically calculates the pest and disease occurrence probability weights of features, completes feature fusion and optimizes feature combinations to achieve post-feature fusion, and finally achieves deep fusion for multi-directional pest and disease occurrence probability prediction; thus obtaining accurate pest and disease images. Attached Figure Description

[0054] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure;

[0055] Figure 1 This is a flowchart of the intelligent identification method for wheat diseases and pests based on meta-learning provided in this embodiment of the invention;

[0056] Figure 2 This is a flowchart of a method for preliminary analysis of wheat diseases and pests based on climate data, soil environmental data, and historical data of the wheat planting area to be identified, provided in an embodiment of the present invention.

[0057] Figure 3 This is a flowchart of a method for obtaining preliminary analysis results of wheat diseases and pests based on historical wheat disease and pest information and predictions of wheat diseases and pests that may occur in the current region, provided by an embodiment of the present invention. Detailed Implementation

[0058] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0059] I. Explanation of the Implementation Example:

[0060] Example 1

[0061] The wheat disease and pest intelligent identification system based on meta-learning provided in this invention includes:

[0062] The environmental data acquisition module, connected to the central control module, is used to collect climate data, soil environmental data, and overall planning data for the wheat planting area to be identified.

[0063] The historical data acquisition module, connected to the central control module, is used to obtain historical pest and disease information and data of wheat in the current planting area from a professional database.

[0064] The sampling point and camera point setting module is connected to the central control module and is used to set the sampling points of soil samples and the placement points of camera equipment based on the overall planning data of the planting area to be identified.

[0065] The image acquisition module, connected to the central control module, is used to acquire images of the leaves and stems of the wheat to be identified using the camera equipment at the placement point;

[0066] The central control module is connected to the environmental data acquisition module, historical data acquisition module, sampling point and camera point setting module, image acquisition module, image preprocessing module, image feature extraction module, pest and disease identification model construction module, meta-learning model construction module, model training module, optimization module, pest and disease preliminary analysis module, soil testing module, pest and disease identification module, and identification result output module. It is used to control the normal operation of each module using a microcontroller or controller.

[0067] The image preprocessing module, connected to the central control module, is used to preprocess the acquired images of the leaves and stems of the wheat to be identified.

[0068] The image feature extraction module, connected to the central control module, is used to extract features from the preprocessed image;

[0069] The pest and disease identification model building module is connected to the central control module and is used to build pest and disease identification models.

[0070] The meta-learning model building module, connected to the central control module, is used to build the meta-learning model for pest and disease identification.

[0071] The model training module, connected to the central control module, is used to train the constructed pest and disease model and the meta-learning model of the pest and disease model using the acquired historical data.

[0072] The optimization module, connected to the central control module, is used to optimize the parameters of the pest and disease identification model using the trained meta-learning model.

[0073] The preliminary analysis module for diseases and pests is connected to the central control module and is used to conduct preliminary analysis of wheat diseases and pests based on the acquired climate data, soil environmental data and historical data of the wheat planting area to be identified.

[0074] The soil testing module, connected to the central control module, is used to collect and test soil data from wheat roots at sampling points.

[0075] The pest and disease identification module is connected to the central control module and is used to identify wheat pests and diseases based on extracted wheat image features using an optimized pest and disease model.

[0076] The identification result output module, connected to the central control module, is used to output wheat disease and pest information based on preliminary identification results, soil testing results, and disease and pest model identification.

[0077] The meta-learning-based intelligent identification system for wheat diseases and pests includes:

[0078] The image preprocessing module, connected to the central control module, is used to preprocess the acquired images of the leaves and stems of the wheat to be identified.

[0079] The image feature extraction module, connected to the central control module, is used to extract features from the preprocessed image;

[0080] The pest and disease identification model building module is connected to the central control module and is used to build pest and disease identification models.

[0081] The meta-learning model building module, connected to the central control module, is used to build the meta-learning model for pest and disease identification.

[0082] The preliminary analysis module for diseases and pests is connected to the central control module and is used to conduct preliminary analysis of wheat diseases and pests based on the acquired climate data, soil environmental data and historical data of the wheat planting area to be identified.

[0083] The soil testing module, connected to the central control module, is used to collect and test soil data from wheat roots at sampling points.

[0084] The pest and disease identification module is connected to the central control module and is used to identify wheat pests and diseases based on extracted wheat image features using an optimized pest and disease model.

[0085] The identification result output module is connected to the central control module and is used to output intelligent identification results of wheat diseases and pests based on the preliminary identification results of wheat diseases and pests, soil test results, and identification of disease and pest models.

[0086] Example 2

[0087] like Figure 1 As shown, the intelligent identification method for wheat diseases and pests based on meta-learning provided in this embodiment of the invention includes:

[0088] S1, The adaptive Wiener filtering method is used to filter the collected images of the leaves and stems of the wheat to be identified;

[0089] S2, perform grayscale processing on the filtered wheat leaf and stem images to obtain grayscale images of wheat leaves and stems;

[0090] S3 uses a crosshair grayscale image sharpness improvement model to enhance the grayscale images of wheat leaves and stems after filtering.

[0091] S4. Based on the enhanced image, identify diseases and pests on wheat leaves and stems, symptoms, lesions, and insect characteristics of pests.

[0092] S5. Based on the obtained climate data, soil environmental data, and historical data of the wheat planting area to be identified, a preliminary analysis of wheat diseases and pests is conducted.

[0093] In this embodiment of the invention, the crosshair grayscale image sharpness improvement model in step S3 is as follows:

[0094] In step S3, the crosshair grayscale image sharpness improvement model is as follows:

[0095]

[0096]

[0097] Where C represents the sharpness of the filtered image; B(I,j) represents the pixel grayscale matrix of the camera at the placement point; m×n represents the number of pixels of the camera at the placement point, 0≤I≤m-1, 0≤J≤n-1, B max B represents the maximum grayscale value of the crosshair grayscale image; min B represents the minimum grayscale value. dif This represents the difference in grayscale values.

[0098] In this embodiment of the invention, in step S4, the disease and pest identification model for identifying diseases and pests, symptoms, lesions, and insect characteristics of wheat leaves and stems includes: an input layer and a feature extraction network.

[0099] The input layer is used to take the extracted wheat feature image as input to the pest and disease identification model;

[0100] The feature extraction network is used to extract information from the input wheat feature image using a residual network that integrates a multi-layer attention mechanism; and the information extracted by the feature extraction network is normalized using a softmax function.

[0101] In this embodiment of the invention, the method for identifying pests and diseases using a pest and disease identification model includes:

[0102] Step 1: Using multiple complementary features, based on different extraction methods, and heterogeneous with each other, we depict images of wheat leaves and stems from different visual perspectives, showing the symptoms, lesions, and insect body characteristics of diseases and pests.

[0103] Step 2: Deeply explore the similar pest and disease characteristics contained among different features to achieve real-time feature fusion;

[0104] Step 3: Dynamically calculate the pest and disease occurrence probability weights of features, complete feature fusion and optimize feature combinations to achieve feature fusion, and finally realize deep fusion of multi-dimensional pest and disease occurrence probability prediction.

[0105] In the first step of this invention, based on the complementarity between heterogeneous features, a set of excellent features of the image are extracted from multiple perspectives such as shape, texture, and deep learning.

[0106] In step 2, the mining models include DCA model, CCL model, DCCA model, SCCA model, MLP model, CCA model, feature concatenation, and multi-kernel learning fusion model;

[0107] The similar pest and disease characteristics include: different characteristics all point to the same or similar pests and diseases, and are represented by SG, SV, SR, GH, GV, GR and SD respectively, for a total of 7 groups of similar pest and disease characteristics;

[0108] In step 3, the post-fusion implementation method specifically includes: using a hybrid soft and hard voting mechanism to vote on the feature combinations generated by the intermediate fusion, thereby achieving feature post-fusion;

[0109] The post-feature fusion implementation method specifically includes:

[0110] SUM fusion is performed, and decision-making is based on Max. Two strategies, random or sorted, are designed. The n′ feature combinations among the top n groups of pest and disease occurrence probability feature combinations with the best accuracy are n=11, n′=3,5,7,9,11. Soft and hard voting decisions are made on the prediction results, and integrated learning of different feature discrimination results is carried out to achieve feature fusion in the later stage.

[0111] The method for achieving feature fusion specifically includes: dynamically calculating feature weights based on the pest and disease identification model to achieve feature fusion;

[0112] The specific methods for implementing the fusion of the features include:

[0113] The Adaboost algorithm is used to calculate the estimated probability of single features and similar pest and disease features; the effective region gene optimization algorithm is improved to dynamically calculate the pest and disease occurrence probability weights of single features and similar pest and disease features, and the probability is estimated by weighting the pest and disease occurrence probability weights to construct feature combinations. The improved pest and disease occurrence probability model is used to achieve fusion of multiple features.

[0114] In the soft and hard voting decision-making process for the prediction results, the decision may adopt a stacked boosting fusion algorithm or a weighted fusion strategy.

[0115] After step 3, the identification of diseases and pests, symptoms, lesions, and insect characteristics of wheat leaves and stems is performed. Based on the identification results, predicted labels for the images of diseases and pests, symptoms, lesions, and insect characteristics of wheat leaves and stems are output.

[0116] Example 3

[0117] like Figure 2 As shown, the preliminary analysis of wheat diseases and pests using the disease and pest analysis module, based on the acquired climate data, soil environmental data, and historical data of the wheat planting area to be identified, includes:

[0118] S201: Obtain historical information on wheat diseases and pests, and analyze climate data and soil environmental data of the wheat planting area to be identified;

[0119] S202, Based on the analysis results of climate data and soil environment data of the wheat planting area, combined with the suitable environment for wheat diseases and pests, predict the diseases and pests that will occur in the wheat in the current area;

[0120] S203, Based on the historical information on wheat diseases and pests and the prediction of wheat diseases and pests that will occur in the current area, a preliminary analysis result of wheat diseases and pests is obtained.

[0121] like Figure 3 As shown, the preliminary analysis results of wheat diseases and pests provided in this embodiment of the invention, based on the historical information on wheat diseases and pests and the prediction of wheat diseases and pests that will occur in the current region, include:

[0122] S301, the first wheat disease and pest is obtained by taking the intersection of historical wheat disease and pest information and the predicted wheat diseases and pests in the current area.

[0123] S302 assigns weights to historical wheat pest and disease information and to the predicted pests and diseases that may occur in wheat in the current region.

[0124] S303, determine the weights of various diseases and pests included in the historical information on wheat diseases and pests and the data on the diseases and pests that will occur in wheat in the current region.

[0125] S304 outputs the analysis results as the primary wheat pest and disease data, along with other pest and disease data arranged in order of weight.

[0126] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0127] The information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of the present invention. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0128] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments.

[0129] II. Application Examples:

[0130] This invention also provides a computer device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above method embodiments.

[0131] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps described in the various method embodiments above.

[0132] This invention also provides an information data processing terminal, which, when executed on an electronic device, provides a user input interface to implement the steps described in the above method embodiments. The information data processing terminal is not limited to mobile phones, computers, or switches.

[0133] This invention also provides a server that, when executed on an electronic device, provides a user input interface to implement the steps described in the above method embodiments.

[0134] This invention provides a computer program product that, when run on an electronic device, enables the electronic device to implement the steps described in the various method embodiments above.

[0135] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0136] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0137] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent identification of wheat diseases and pests based on meta-learning, characterized in that, The method includes the following steps: S1, The adaptive Wiener filtering method is used to filter the collected images of the leaves and stems of the wheat to be identified; S2, perform grayscale processing on the filtered wheat leaf and stem images to obtain grayscale images of wheat leaves and stems; S3 uses a crosshair grayscale image sharpness improvement model to enhance the grayscale images of wheat leaves and stems after filtering. S4. Based on the enhanced image, identify diseases and pests on wheat leaves and stems, symptoms, lesions, and insect characteristics of pests. S5. Based on the obtained climate data, soil environmental data and historical data of the wheat planting area to be identified, a preliminary analysis of wheat diseases and pests is conducted. In step S4, the pest and disease identification model for identifying diseases and pests, symptoms, lesions, and insect characteristics of wheat leaves and stems includes: an input layer and a feature extraction network. The input layer is used to take the extracted wheat feature image as input to the pest and disease identification model; The feature extraction network is used to extract information from the input wheat feature image using a residual network that integrates a multi-layer attention mechanism; and the information extracted by the feature extraction network is normalized using a softmax function. The identification methods for pest and disease identification models include: Step 1: Using multiple complementary features, based on different extraction methods, and heterogeneous with each other, we depict images of wheat leaves and stems from different visual perspectives, showing the symptoms, lesions, and insect body characteristics of diseases and pests. Step 2: Deeply explore the similar pest and disease characteristics contained among different features to achieve real-time feature fusion; Step 3: Dynamically calculate the pest and disease occurrence probability weights of features, complete feature fusion and optimize feature combinations to achieve feature fusion, and finally realize deep fusion of multi-dimensional pest and disease occurrence probability prediction. Step 4 involves identifying images of diseases and pests, symptoms, lesions, and insect characteristics on wheat leaves and stems, and outputting predicted labels based on the identification results.

2. The intelligent identification method for wheat diseases and pests based on meta-learning according to claim 1, characterized in that, In step S3, the crosshair grayscale image sharpness improvement model is as follows: ； ； Where C represents the sharpness of the filtered image; The pixel grayscale value matrix of the camera device at the placement point; The pixel count of the camera device at the placement point. B max B represents the maximum grayscale value of the crosshair grayscale image; min B represents the minimum grayscale value. dif This represents the difference in grayscale values.

3. The intelligent identification method for wheat diseases and pests based on meta-learning according to claim 1, characterized in that, In step 1, the heterogeneity based on different extraction methods includes: extracting a set of excellent features from multiple perspectives such as shape, texture, and deep learning based on the complementarity between heterogeneous features; In step 2, the in-depth mining of similar pest and disease features contained in different features includes: mining similar pest and disease features through DCA model, CCL model, DCCA model, SCCA model, MLP model, CCA model, feature splicing, and multi-kernel learning fusion model; among them, similar pest and disease features include: different features all point to the same or similar pests and diseases, and are represented by SG, SV, SR, GH, GV, GR, and SD respectively, for a total of 7 groups of similar pest and disease features; In step 3, the method for achieving feature fusion and optimal feature combination after feature integration specifically includes: using a hybrid soft and hard voting mechanism to vote on the feature combinations generated by feature integration, performing SUM fusion, and simultaneously making decisions based on Max. Two strategies are designed: random or sorted. The top n groups of pest and disease occurrence probability feature combinations with the highest accuracy are selected. Group feature combination, The prediction results are evaluated using both soft and hard voting methods, and integrated learning is performed on the results of different feature discriminations to achieve post-feature fusion.

4. The intelligent identification method for wheat diseases and pests based on meta-learning according to claim 1, characterized in that, In step 3, the method for implementing the fusion specifically includes: calculating the estimated probability of single features and similar pest and disease features based on the Adaboost algorithm; improving the effective region gene optimization algorithm, dynamically calculating the pest and disease occurrence probability weights of single features and similar pest and disease features, using the pest and disease occurrence probability weights to estimate the probability, constructing feature combinations, and using the improved pest and disease occurrence probability model to achieve fusion of multiple features. In the process of making soft or hard voting decisions on the prediction results, the decision may employ a stacked boosting fusion algorithm or a weighted fusion strategy.

5. The intelligent identification method for wheat diseases and pests based on meta-learning according to claim 1, characterized in that, The preliminary analysis of wheat diseases and pests specifically includes the following steps: First, historical information on wheat diseases and pests was obtained, and climate and soil environmental data of the wheat planting areas to be identified were analyzed. Secondly, based on the analysis results of climate data and soil environment data of the wheat planting area, combined with the suitable environment for wheat diseases and pests, the diseases and pests that may occur in the wheat in the current area are predicted. Finally, based on the historical information on wheat diseases and pests and the prediction of wheat diseases and pests that may occur in the current region, preliminary analysis results on wheat diseases and pests were obtained.

6. The intelligent identification method for wheat diseases and pests based on meta-learning according to claim 5, characterized in that, The preliminary analysis results of wheat diseases and pests specifically include: (1) The first wheat disease and pest is obtained by taking the intersection of the historical wheat disease and pest information and the predicted wheat diseases and pests in the current area. (2) Assign weights to the historical pest and disease information of wheat and the predicted pest and disease outbreaks in the current region; (3) Determine the weights of various diseases and pests included in the historical information on wheat diseases and pests and the data on diseases and pests that will occur in wheat in the current region. (4) Output the data of the first wheat disease and pest and the other diseases and pests arranged in order of weight as the analysis results.

7. A meta-learning-based intelligent identification system for wheat diseases and pests, used to implement the meta-learning-based intelligent identification method for wheat diseases and pests as described in any one of claims 1-6, characterized in that, The meta-learning-based intelligent identification system for wheat diseases and pests includes: The central control module uses a microcontroller or controller to control the normal operation of each module. The image preprocessing module, connected to the central control module, is used to preprocess the acquired images of the leaves and stems of the wheat to be identified. The image feature extraction module, connected to the central control module, is used to extract features from the preprocessed image; The pest and disease identification model building module is connected to the central control module and is used to build pest and disease identification models. The meta-learning model building module, connected to the central control module, is used to build the meta-learning model for pest and disease identification. The preliminary analysis module for diseases and pests is connected to the central control module and is used to conduct preliminary analysis of wheat diseases and pests based on the acquired climate data, soil environmental data and historical data of the wheat planting area to be identified. The soil testing module, connected to the central control module, is used to collect and test soil data from wheat roots at sampling points. The pest and disease identification module is connected to the central control module and is used to identify wheat pests and diseases based on extracted wheat image features using an optimized pest and disease model. The identification result output module is connected to the central control module and is used to output intelligent identification results of wheat diseases and pests based on the preliminary identification results of wheat diseases and pests, soil test results, and identification of disease and pest models.

8. The intelligent identification system for wheat diseases and pests based on meta-learning according to claim 7, characterized in that, The meta-learning-based intelligent identification system for wheat diseases and pests also includes: The environmental data acquisition module, connected to the central control module, is used to collect climate data, soil environmental data, and overall planning data for the wheat planting area to be identified. The historical data acquisition module, connected to the central control module, is used to obtain historical pest and disease information and data of wheat in the current planting area from a professional database. The sampling point and camera point setting module is connected to the central control module and is used to set the sampling points of soil samples and the placement points of camera equipment based on the overall planning data of the planting area to be identified. The image acquisition module, connected to the central control module, is used to acquire images of the leaves and stems of the wheat to be identified using the camera equipment at the placement point; The model training module, connected to the central control module, is used to train the constructed pest and disease model and the meta-learning model of the pest and disease model using the acquired historical data. The optimization module, connected to the central control module, is used to optimize the parameters of the pest and disease identification model using the trained meta-learning model.

Images