Crop disease and pest identification and analysis method based on image processing

Abstract

The invention belongs to the technical field of crop disease and insect pest recognition, and particularly discloses a crop disease and insect pest recognition analysis method based on image processing, and the method comprises the steps: collecting a state image of a plant in a dynamic environment, and eliminating the interference of the dynamic environment on the state image recognition of a plant leaf; differentiated analysis is carried out on disease and insect pest characteristics of different areas corresponding to each leaf area, so that leaf disease and insect pest types and defect data thereof are accurately identified; the color value distribution health state of leaves near a leaf vein distribution area is detected by simulating the leaf vein distribution contour of the interference area, the corresponding leaf vein structure stability of each leaf area is analyzed, and the leaf state of the remaining area after the interference area and the pest and disease damage area of the leaves are removed is identified. And assisting in evaluating the leaf risk area condition in each leaf area, and judging the pest and disease damage state in the plant leaf according to the condition. The recognition precision of the crop state image is improved, and targeted prevention and control measures can be taken.

Description

Technical Field

[0001] This invention belongs to the field of crop pest and disease identification technology, and relates to a crop pest and disease identification and analysis method based on image processing. Background Technology

[0002] Crop diseases and pests are a significant factor affecting agricultural production, posing a serious threat to food security and economic development for a long time. With the development of information technology, computer vision technology has been gradually applied to the field of crop disease and pest identification. However, traditional image processing methods are easily affected by factors such as dynamic environments when faced with complex disease and pest images, making it difficult to accurately identify disease and pest characteristics, thus resulting in insufficient accuracy and efficiency in crop disease and pest identification.

[0003] In existing technologies, there are also some solutions related to the identification of crop diseases and pests. For example, Chinese Patent Publication No. CN113763304A discloses a method, device, equipment, and medium for identifying crop diseases and pests. This method involves acquiring a first crop image; separating the RGB channels of the first crop image to obtain R-channel, G-channel, and B-channel images; binarizing the R-channel, G-channel, and B-channel images to obtain R-channel binary images, G-channel binary images, and B-channel binary images; performing block blob analysis based on the R-channel, G-channel, and B-channel binary images to determine the location of diseases and pests and generate disease and pest data; and inputting the disease and pest data into a decision tree model to determine the type of disease and pest. This approach can quickly determine the location and type of diseases and pests, improving the accuracy of identification.

[0004] Chinese Patent Publication No. CN110544237B discloses a training and recognition method for a camellia oleifera pest and disease model based on image analysis. This method involves obtaining leaf images containing original pest and disease information; segmenting the leaf images to retain lesion areas, thus accurately obtaining images of these lesion areas. This segmentation can be based on the type of pest or disease, or on different stages of the same pest or disease. The segmented images are then labeled and submitted to a neural network for training; this neural network is then used for pest and disease identification. Further precise segmentation of the leaf images containing pests and diseases allows the neural network to extract more accurate color features, texture features, and morphological information, resulting in higher accuracy and significantly promoting the scientific and accurate control of pests and diseases.

[0005] Although the above-mentioned solutions propose some methods for identifying crop diseases and pests, they still have the following limitations: 1. Existing identification methods are usually based on image acquisition under normal conditions, and lack investigation of dynamic environmental interference factors in the crop growing area, which may lead to misleading results in the acquisition and analysis of crop status images.

[0006] 2. Current methods for locating leaf disease and pest characteristics tend to focus on surface defects and textures, lacking analysis of vein structure texture stability. Fine cracks and texture interruptions on veins can be key signals of disease and pest invasion. The impact of diseases and pests on veins leads to abnormal chlorophyll distribution, affecting nutrient transport processes. However, these issues are easily overlooked in the overall analysis, resulting in inaccurate disease and pest location. Therefore, without analysis of vein texture stability, a deeper understanding of the mechanisms by which diseases and pests affect leaf physiological functions may be lacking, preventing a fundamental solution to the disease and pest problem. Summary of the Invention

[0007] In view of this, in order to solve the problems mentioned in the background technology, a method for identifying and analyzing crop diseases and pests based on image processing is proposed.

[0008] The objective of this invention can be achieved through the following technical solution: This invention provides a method for identifying and analyzing crop diseases and pests based on image processing. The method includes the following steps: Step 1: Acquire images of crop plants under the current dynamic environment, including haze, rain, and snow. Mark each leaf region in the crop plant image and detect relevant meteorological data related to the current dynamic environment of the crop. Each leaf region is numbered as follows: .

[0009] Step 2: Identify the interference factors under various dynamic environments and delineate the interference areas of each leaf region in the crop plant image corresponding to the dynamic environment.

[0010] Step 3: Extract historical images of crop diseases and pests, delineate the corresponding disease and pest areas in each leaf region of the crop plant image, and identify the types of diseases and pests, including lesion type, insect damage type, and curling type.

[0011] Step 4: Integrate the identification regions within each leaf area of ​​the crop plant image, and obtain effective information on the corresponding disease and pest areas and identification regions within each leaf area of ​​the crop plant image.

[0012] Step 5: Based on valid information, identify leaf analysis results for each leaf region in the crop plant image, including the leaf risk area index of the crop plant. Canopy coverage This allows for the assessment of crop plant disease and pest indices.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention collects the state images of the plant in the dynamic environment, eliminates the interference of the dynamic environment on the identification of the plant leaf state images, and judges the state of pests and diseases in the plant leaves, which increases the recognition accuracy of crop state images, and makes the identification results of the corresponding types and ranges of crop pests and diseases more reliable.

[0014] (2) By conducting differential analysis of the pest and disease characteristics of different areas of each leaf, this invention can accurately identify the types of pests and diseases on the leaves and their defect data. Furthermore, when identifying pest and disease types, the main pest and disease types are weighted, which can highlight the key pest and disease types that have a greater impact on crops and a wider range of influence. Based on this, the leaf risk area index of crop plants can be evaluated, which helps to take targeted prevention and control measures and improve the prevention and control effect.

[0015] (3) This invention simulates the leaf vein distribution outline of the interference area and detects the health status of the color value distribution of leaves near the leaf vein distribution area. It analyzes the stability of the corresponding leaf vein structure in each leaf area and helps to assess the leaf risk area in each leaf area by identifying the leaf status of the remaining area after removing the interference area and the disease and pest area. Changes in the color value of leaves around the leaf vein are often an early signal of leaf disease. Many diseases will first cause color changes in the area around the leaf vein in the early stage. By timely monitoring these color value changes, problems can be detected before the disease spreads on a large scale, thus providing an opportunity for early prevention and control and effectively controlling the spread of the disease. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the implementation steps of the method of the present invention.

[0018] Figure 2 This is a schematic diagram of the corresponding disease and pest areas in the leaf area of ​​this invention.

[0019] Figure 3 This is a complete outline view of the leaf vein distribution in the leaf area of ​​the present invention.

[0020] Attached labels: 1. Area affected by pests and diseases; 2. Veins of each branch. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 As shown, this invention provides a method for identifying and analyzing crop diseases and pests based on image processing. The method includes the following steps: Step 1: Acquire images of crop plants under dynamic environments, including hazy, rainy, and snowy conditions. Obtain the outer contour features of the crop plant stems and leaves. Based on contour recognition technology, mark each leaf region in the crop plant image under dynamic environments and detect relevant meteorological data of the current dynamic environment. Each leaf region is numbered as follows: .

[0023] The crop plant image refers to an image of the entire stem and leaves of a single crop plant.

[0024] Step 2: Identify the interference factors under various dynamic environments, construct a set of interference factors for each dynamic environment, and delineate the interference regions for each leaf area in the crop plant image under the corresponding dynamic environment.

[0025] In a preferred embodiment, the dynamic environment in which the crop is currently located is related to meteorological data such as air visibility in a hazy environment, rainfall rate in a rainy environment, and snowfall rate in a snowy environment.

[0026] Interference factors in various dynamic environments include air blurring factors in hazy environments, raindrop interference factors in rainy environments, and snowflake interference factors in snowy environments.

[0027] In a further preferred embodiment, the step of delineating the interference regions of the dynamic environment corresponding to each leaf region in the crop plant image includes: extracting the interference factor map set corresponding to the dynamic environment in which the crop is currently located from the interference factor map set corresponding to each dynamic environment, and identifying the contour texture of each sub-map in the interference factor map set corresponding to the dynamic environment in which the crop is currently located based on contour recognition technology.

[0028] The crop plant image is divided into sub-regions according to the grid division method. Then, the leaf surface filamentous texture of each leaf region in the crop plant image is identified. This is compared with the contour texture of each sub-image in the corresponding interference factor map set of the current dynamic environment of the crop. The sub-regions that match the leaf surface filamentous texture of each leaf region in the crop plant image with the contour texture of each sub-image in the corresponding interference factor map set of the current dynamic environment of the crop are identified and recorded as the interference regions of the corresponding dynamic environment of each leaf region in the crop plant image.

[0029] The contour textures include edge gloss contrast textures and internal reflective stripe textures of raindrops, white patch textures of snowflakes, and surface bump textures.

[0030] The leaf surface has filamentous textures such as vein morphology and color characteristics.

[0031] Specifically, the outline of raindrops on a leaf is usually circular or elliptical with relatively smooth edges, forming a clear boundary with the leaf surface, resulting in a significant gloss contrast between the raindrop and the leaf. Due to the transparency and surface tension of raindrops, some fine textures can be observed inside the raindrop in high-resolution images. This is caused by the refraction and reflection of light, as well as tiny impurities or bubbles inside the raindrop, which may result in certain striped textures inside the raindrop.

[0032] Snowflakes typically accumulate on leaves, with multiple snowflakes merging together. During the accumulation process, the boundaries between snowflakes gradually blur, forming irregular white patches. The edges of these patches may be uneven, and they contain the outlines and structures of multiple snowflakes, presenting a rich sense of layering and three-dimensionality. Furthermore, their surfaces have some tiny textures and unevenness, manifested as fine lines, particles, or bumps on the surface of the snowflake.

[0033] Please see Figure 2 As shown, step three involves extracting historical images of crop diseases and pests, constructing a leaf atlas of crop diseases and pests, combining it with relevant meteorological data on the current dynamic environment of the crop, delineating the corresponding disease and pest areas in each leaf region of the crop plant image, and identifying the types of diseases and pests, including lesion type, insect damage type, and curling type.

[0034] In a preferred embodiment, the step of delineating each leaf region in the crop plant image and corresponding to each pest and disease region, and identifying the pest and disease type, includes: constructing a crop pest and disease leaf atlas based on historical crop pest and disease images, which includes various pest and disease types and corresponding pest and disease defect textures, such as defect textures corresponding to lesion types, insect-eaten types, and curled-up types.

[0035] By taking the inverse regions corresponding to the interference regions of the dynamic environment of each leaf region in the crop plant image, the remaining sub-intervals of each leaf region in the crop plant image are obtained. The leaf filamentous texture of each remaining sub-interval of each leaf region in the crop plant image is statistically analyzed, and its image training model is constructed. The meteorological data related to the current dynamic environment of the crop are substituted into its image training model. Based on the physical model algorithm, the leaf filamentous texture of each remaining sub-interval of each leaf region is obtained as a result of the training.

[0036] It should be noted that the corresponding reverse region of each interference region in the dynamic environment of each blade region is: the comprehensive remaining region of each blade region excluding each interference region.

[0037] Based on contour recognition technology, the disease and pest textures of each sub-image in the crop disease and pest leaf image set are extracted. Then, they are compared with the leaf surface filamentous textures of the corresponding remaining sub-intervals of each leaf region in the training results. The remaining sub-intervals that match the leaf surface filamentous textures of each leaf region in the crop plant image with the disease and pest textures of each sub-image in the crop disease and pest leaf image set are identified and recorded as the corresponding disease and pest regions of each leaf region in the crop plant image. The disease and pest types to which the disease and pest textures of these matching sub-images belong are obtained, thus obtaining the disease and pest types of each disease and pest region of each leaf region in the crop plant image.

[0038] It should be noted that the specific content of the physical model algorithm is as follows: For hazy environments, based on the air visibility of the hazy environment, an appropriate image enhancement precision is matched, and enhancement training is performed on each sub-interval of each leaf region to make the leaf surface filamentous texture of each sub-interval of each leaf region clearer; For rainy environments, based on the rainfall speed of the rainy environment, the corresponding raindrop change shape (such as raindrop volume size) is matched; For snowy environments, based on the snowfall speed of the snowy environment, the corresponding snow-covered leaf shape (such as snowflakes melting into water or not melting) is matched.

[0039] This invention collects images of plants in a dynamic environment, eliminates interference from the dynamic environment on the identification of plant leaf images, and determines the status of pests and diseases in plant leaves. This increases the accuracy of crop status image recognition, thereby making the identification results of the corresponding types and ranges of crop pests and diseases more reliable.

[0040] Step 4: Integrate the identification regions within each leaf area of ​​the crop plant image, and obtain effective information on the corresponding disease and pest areas and identification regions within each leaf area of ​​the crop plant image.

[0041] The effective information in the crop plant image for each leaf region corresponding to each pest and disease region and identification region includes: the pest and disease type and defect data of each pest and disease region in each leaf region, and the leaf surface filamentous texture corresponding to the identification region of each leaf region.

[0042] The leaf surface filamentous texture corresponding to the identification area of ​​each leaf region is obtained by screening the leaf surface filamentous texture of each sub-interval of each leaf region.

[0043] In a preferred embodiment, the integration of the identification regions within each leaf region of the crop plant image specifically involves summarizing the corresponding remaining sub-intervals of each leaf region in the crop plant image to obtain the comprehensive remaining region of each leaf region in the crop plant image.

[0044] In the comprehensive remaining area of ​​each leaf region in the crop plant image, the corresponding inverse region of each disease and pest region is taken to construct the identification region of each leaf region in the crop plant image.

[0045] It should be noted that the corresponding reverse areas of each disease and pest area are: the comprehensive remaining areas of each leaf area excluding the comprehensive disease and pest areas.

[0046] Step 5: Based on valid information, identify leaf analysis results for each leaf region in the crop plant image, including the leaf risk area index of the crop plant. Canopy coverage Then, the pest and disease index of crop plants is assessed, and then, by combining the above analysis steps, all plants in the planting area of ​​the crop plants are analyzed one by one.

[0047] In a preferred embodiment, the leaf risk area index of the crop plant is obtained by: determining the pest and disease prevalence type of each leaf area based on the pest and disease type of each leaf area corresponding to each pest and disease area in the crop plant image, and setting the pest and disease threat factor for each leaf area accordingly. , This indicates the number of each pest and disease affected area. .

[0048] Defect data corresponding to the pest and disease types in each leaf region are extracted from effective information and imported into the leaf risk area index assessment model for the corresponding pest and disease type. For example, for insect-eaten defect textures, the area of ​​insect holes is extracted; for lesion textures, the area of ​​lesions is extracted; and for curled leaf textures, the area and degree of leaf curling are extracted. A leaf risk area index assessment model is then constructed. Based on this method, the leaf risk area index for each disease and pest region in each leaf area was derived. .

[0049] In the formula, This represents the leaf risk area index, indicating the area of ​​disease and pest infestation within the leaf region. This indicates a preset reference area. This indicates the types of pests and diseases affecting the affected area. The types of pests and diseases in the affected areas are respectively: insect-eaten type, lesion type, and curled-up type. These represent the defect areas of the affected areas corresponding to insect-eaten, lesion, and curling types, respectively. Indicates pest and disease threat factors in pest and disease-affected areas. This indicates the degree of leaf curling in the affected area corresponding to the type of curling.

[0050] It should be noted that the degree of blade curling refers to the ratio of the curling curvature of the blade curling profile to the preset reference curling curvature.

[0051] Based on the fundamental differences in the degree of damage to crop leaves caused by different types of pests and diseases, and the fact that leaf size heterogeneity makes direct comparison of the defect area of ​​pests and diseases meaningless, and considering the actual planting scenario that the pests and diseases with the highest proportion in leaves pose a significantly higher threat to plant growth than other types, and also considering that the degree of curling of pests and diseases needs to be additionally correlated to accurately represent the state of damage, a segmented leaf risk area index assessment model was constructed.

[0052] This leaf risk area index assessment model specifically addresses the problem of existing technologies relying solely on leaf surface defect texture for location, failing to consider differences in pest and disease types and leaf size heterogeneity, leading to distorted risk quantification. By normalizing defect area, it makes the risk of pest and disease areas on leaves of different sizes comparable; by weighting threat factors, it highlights the impact of major pest and disease types, preventing secondary types from masking core risks, thereby providing reliable basic data for subsequent overall leaf risk calculation and improving the accuracy of pest and disease risk quantification.

[0053] Extract the corresponding leaf surface filamentous texture of the identification region in each leaf area from the effective information of crop plant images, and analyze the leaf risk area index of the identification region in each leaf area. This leads to the construction of a leaf risk area index for crop plants. , where e is the natural constant.

[0054] In a further preferred embodiment, determining the pest and disease prevalence type of each leaf region and setting the pest and disease threat factor for each leaf region accordingly includes: obtaining the area of ​​each pest and disease region corresponding to each leaf region in the crop plant image; summarizing the total area of ​​the same pest and disease type in each leaf region of the crop plant image based on the pest and disease type in each leaf region of the crop plant image; obtaining its proportion relative to the total area of ​​the corresponding leaf region in the crop plant image; and selecting the pest and disease type with the largest proportion in each leaf region of the crop plant image as the pest and disease prevalence type for each leaf region.

[0055] Match the pest and disease types of each leaf region with the pest and disease prevalence type of the corresponding leaf region. If the pest and disease type of a certain leaf region matches the pest and disease prevalence type of the corresponding leaf region, then record the pest and disease threat factor of that leaf region as follows: Conversely, the pest threat factor for that leaf area corresponding to the pest-affected area is recorded as... , Statistics on pest and disease threat factors in each leaf area and corresponding pest and disease areas were compiled. , .

[0056] Please see Figure 3 As shown, in a further preferred embodiment, the analysis of the leaf risk area index of the identification area in each leaf region includes: Q1, obtaining the leaf filamentous texture of each interference area in the dynamic environment of each leaf region in the crop plant image, combining the effective information of each disease and pest area and identification area in each leaf region in the crop plant image, constructing the contour information of each interference area in the dynamic environment of each leaf region through geometric shape fitting, and integrating it into the complete contour of the leaf vein distribution of each leaf region in the crop plant image.

[0057] It should be noted that the geometric shape fitting method, such as polygon fitting, is applicable when simulating the contour information of interference areas on the blade surface. For example, when simulating the contour information of interference areas on the blade surface, polygon fitting is suitable for situations where it is necessary to simplify the representation of the texture or extract the main contour features. For instance, for blades with complex textures, polygon fitting can extract the approximate shape of the blade or the polygonal contour formed by the main vein branches, facilitating rapid analysis of the overall shape and main texture direction of the blade.

[0058] Q2. Identify the complete outline of the vein distribution in each leaf region and the thickness of the corresponding vein branches. and texture interruption ratio Mark the sub-regions adjacent to the veins of each branch of each leaf region, and extract their region color values, denoted as . , Indicates the numbering of the veins in each branch. , This indicates the sub-region number adjacent to the branch vein. .

[0059] The percentage of texture interruption in the branch veins refers to the proportion of the area of ​​texture interruption caused by crack defects or unclear veins in the leaf relative to the total area of ​​the branch veins.

[0060] The color value refers to the color coordinate value of the green channel in the RGB color channel.

[0061] Q3. Obtain the standard leaf vein thickness of healthy crop leaves. Standard healthy color value Standard thickness Based on this, the stability of the corresponding vein structure in each leaf region was analyzed. ,in These represent the preset influence weights of leaf vein thickness and healthy color value, respectively. .

[0062] Specifically, the leaf veins are chlorophyll donors for plant leaves, and the chlorophyll content in the adjacent areas is even higher. Therefore, when the color value of the adjacent areas of the leaf veins does not meet the standard or the thickness of the leaf veins does not meet the standard, it indicates that there is internal damage to the overall vein structure of the leaf.

[0063] The construction of the formula for calculating the stability of leaf vein structure in each leaf region can effectively solve the problem that existing technologies lack analysis of the stability of leaf vein structure texture and only focus on leaf surface defects, leading to the omission of potential damage to leaf physiological functions by early pests and diseases. Through the comprehensive evaluation of leaf vein thickness and adjacent color values, the hidden damage of pests and diseases to leaf vein structure can be accurately identified, abnormal leaf physiological functions can be detected in advance, data support can be provided for early prevention and control, and the spread of pests and diseases caused by the omission of early risks in existing technologies can be avoided.

[0064] Q4. Construct a grid-based sub-interval for the identification region within each leaf region, denoted as the sub-interval of the identification region within each leaf region. Then, based on contour recognition technology, identify the leaf color value of each sub-interval of the identification region within each leaf region from the corresponding leaf surface filamentous texture. Blade thickness Assess the color brightness of the distinguishing regions within each leaf area. ,in Indicates the first The identification area within the leaf region of the first leaf region The leaf color value of each sub-interval This indicates the difference in preset reference color values. This indicates the sub-interval numbers of the identification region. , This indicates the number of sub-intervals in the region to be identified.

[0065] Specifically, the grid division of the identification region within each leaf area is as follows: after dividing each leaf area in the crop plant image into sub-intervals according to the grid division method, the remaining sub-intervals obtained after screening are excluding each interference area and each pest and disease area.

[0066] Q5. Analyze the leaf risk area index of the identification region in each leaf region. .

[0067] By weighted fusion of leaf vein stability and color brightness, the system comprehensively covers both visible and hidden risk scenarios in leaves, ensuring that leaf risk assessment is thorough and provides complete leaf risk data for subsequent plant-level risk assessment, thereby enhancing the comprehensiveness of risk assessment.

[0068] This invention simulates the leaf vein distribution outline in the interference area and detects the health status of the color value distribution of leaves near the vein distribution area. It analyzes the stability of the corresponding leaf vein structure in each leaf region and, by identifying the leaf condition of the remaining areas after removing interference and pest / disease-affected areas, assists in assessing the risk area of ​​leaves in each region. Changes in leaf color value around the veins are often early signals of leaf disease. Many diseases initially cause color changes in the area around the veins. For example, when leaves are infected with fungal diseases, small pale yellow or light brown spots may appear around the veins. This is because after the pathogen invades the leaf, it first affects the cell metabolism around the veins. By timely monitoring these color value changes, problems can be detected before the disease spreads extensively, thus providing an opportunity for early prevention and control, effectively controlling the spread of the disease.

[0069] In a further preferred embodiment, the canopy coverage is obtained by integrating the total area of ​​each leaf region. And obtain the total area of ​​the identification region for each leaf region. and the total area of ​​pest and disease areas The effective area ratio of each blade region was obtained by comparison. .

[0070] Count the total number of leaf regions in crop plant images And identify the size characteristics of crop plant stems. Analyze the canopy coverage of crop plants ,in This indicates the weight of the total number of leaf regions relative to the size characteristics of crop stems.

[0071] The calculation of crop canopy coverage is mainly based on the existence of interference areas in crop plant images under dynamic environments. The proportion of effective detection areas directly affects the reliability of pest and disease analysis data. The size of plant stems can characterize the rationality of leaf growth distribution, and thus, the quality of effective detection areas is taken into consideration. Therefore, the canopy coverage of crop plants is constructed by combining the average proportion of effective areas of a single leaf with the total number of leaf areas and the characteristics of crop plant stem size.

[0072] It can specifically address the problems of existing technologies, such as the lack of investigation of dynamic environmental interference and the failure to define effective detection areas, which leads to misleading analysis results. By calculating the proportion of effective areas and stem size characteristics, it accurately maps the reliability of image data, provides a weighting basis for subsequent plant-level risk assessment, avoids invalid area data from interfering with analysis results, and improves the credibility of pest and disease analysis data.

[0073] The dimensions of the crop plant stems include the height and thickness of the stems. These dimensions are used as numerators and compared with preset reference height and reference thickness, respectively. The results are then accumulated and mapped to the characteristics of the crop plant stem dimensions.

[0074] In a further preferred embodiment, the evaluation formula for assessing the crop plant disease and pest index is as follows: .

[0075] It should be noted that the greater the canopy coverage of the crop plants, the more identifiable areas there are in the crop images, and thus the greater the reliability of the crop disease and pest analysis data.

[0076] This invention analyzes the canopy coverage of crop plants to map the proportion of the effective detection area in the crop images relative to the total area of ​​the plant. It can also identify the density of leaf growth in the entire crop plant, thereby increasing the accuracy of crop plant disease and pest index assessment.

[0077] This invention, through differential analysis of the pest and disease characteristics of different areas of each leaf, can accurately identify the types of pests and diseases affecting leaves and their defect data. Furthermore, when identifying pest and disease types, a weighted setting is applied to the main pest and disease types. Different pests and diseases have different degrees of damage to crop leaves. By weighting, the key pest and disease types that cause greater damage to crops and have a wider impact can be highlighted. Based on this, the leaf risk area index of crop plants can be assessed, which helps to take targeted prevention and control measures and improve the prevention and control effect.

[0078] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for identifying and analyzing crop diseases and pests based on image processing, characterized in that, include: The system collects images of crop plants under dynamic environments, including haze, rain, and snow. It marks the leaf regions in the crop plant images and detects meteorological data related to the current dynamic environment of the crops. Identify the interference factors under various dynamic environments and delineate the interference regions in the corresponding dynamic environments for each leaf area in the crop plant image. Extract historical images of crop diseases and pests, delineate the corresponding disease and pest areas in each leaf region of the crop plant image, and identify the types of diseases and pests, including lesion type, insect damage type, and curling type. The identification regions within each leaf area in the crop plant image are integrated, and effective information on the corresponding disease and pest areas and identification regions in each leaf area of ​​the crop plant image is obtained. Based on valid information, leaf analysis results are used to identify leaf regions in crop plant images, including the leaf risk area index of crop plants. Canopy coverage This allows for the assessment of crop plant disease and pest indices.

2. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 1, characterized in that, The meteorological data related to the current dynamic environment of the crops include air visibility in hazy environments, rainfall rate in rainy environments, and snowfall rate in snowy environments. The interference factors in each dynamic environment include air blurring factors in hazy environments, raindrop interference factors in rainy environments, and snowflake interference factors in snowy environments.

3. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 1, characterized in that, The process of delineating the interference regions corresponding to the dynamic environment of each leaf area in the crop plant image includes: Extract the atlas of disturbance factors corresponding to the current dynamic environment of crops; Based on contour recognition technology, the contour texture of each sub-image in the dynamic environment of the crop is currently in is identified. The crop plant image is divided into sub-intervals for each leaf region according to the grid division method; Identify the leaf surface filamentous texture in each sub-region and compare it with the outline texture of each sub-image; Each sub-region that matches the leaf surface filamentous texture of the leaf region with the outline texture of each sub-image is denoted as the interference region of the corresponding dynamic environment.

4. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 3, characterized in that, The process involves delineating the corresponding disease and pest areas within each leaf region of the crop plant image and identifying the types of diseases and pests. This includes: Based on historical images of crop diseases and pests, construct a leaf atlas of crop diseases and pests. The corresponding interference regions of each leaf region in the dynamic environment of the crop plant image are taken as the inverse regions, and the corresponding remaining sub-intervals of each leaf region in the crop plant image are obtained. The leaf surface filamentous texture of each leaf region in the crop plant image is counted, and its image training model is constructed. By substituting meteorological data related to the current dynamic environment of crops into their image training model, and based on the physical model algorithm, the leaf surface filamentous texture of each leaf region corresponding to each remaining sub-interval training result is obtained. Extract the disease and pest textures from each sub-image in the crop disease and pest leaf image set, and then compare them with the leaf filamentous textures from the training results of each remaining sub-interval in each leaf region. The remaining sub-intervals that match the leaf filamentous texture of each leaf region in the crop plant image with the disease and pest texture of each sub-image in the crop disease and pest leaf image set are recorded as the corresponding disease and pest regions of each leaf region in the crop plant image. Obtain the pest and disease type to which the pest and disease textures of these matching sub-images belong, and obtain the pest and disease type of each leaf region in the crop plant image corresponding to each pest and disease region.

5. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 1, characterized in that, The specific identification regions within each leaf region of the integrated crop plant image are as follows: By summarizing the remaining sub-intervals of each leaf region in the crop plant image, the comprehensive remaining region of each leaf region in the crop plant image is obtained. In the comprehensive remaining area of ​​each leaf region in the crop plant image, the corresponding inverse region of each disease and pest region is taken to construct the identification region of each leaf region in the crop plant image.

6. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 1, characterized in that, The leaf risk area index of the crop plants is obtained as follows: Based on the pest and disease types in each leaf region corresponding to each pest and disease region in the crop plant image, determine the pest and disease prevalence type in each leaf region, and set the pest and disease threat factor in each leaf region corresponding to each pest and disease region accordingly. Extract the defect data of the corresponding defect texture of each disease and pest area in each leaf region, and import it into the leaf risk area index assessment model of the corresponding disease and pest type. For example, for the defect texture of the insect-eaten type, extract the area of ​​the insect hole; for the defect texture of the lesion type, extract the area of ​​the lesion; for the defect texture of the curled type, extract the area of ​​the curled leaf and the degree of leaf curling. Construct the leaf risk area index assessment model, and determine the leaf risk area index of each disease and pest area in each leaf region according to the leaf risk area index assessment model. The leaf risk area index of the identification region in each leaf region of the crop plant image is obtained, and the leaf risk area index of the identification region in each leaf region is analyzed. Then, the leaf risk area index of the crop plant is calculated and determined.

7. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 6, characterized in that, The process of determining the prevalence of pests and diseases in each leaf region, and accordingly setting pest and disease threat factors for each leaf region, includes: Summarize the total area of ​​the same type of pest or disease in each leaf region of the crop plant image. Obtain its proportion relative to the total area of ​​the corresponding leaf region in the crop plant image; The types of pests and diseases with the largest proportion in each leaf area of ​​crop plant images were selected and identified as the pest and disease predominance types in each leaf area. Match the pest and disease types of each leaf region with the pest and disease prevalence type of the corresponding leaf region. If the pest and disease type of a certain leaf area matches the pest and disease prevalence type of the corresponding leaf area, then the pest and disease threat factor of that leaf area corresponding to the pest and disease area is recorded as follows: Conversely, the pest threat factor for that leaf area corresponding to the pest-affected area is recorded as... ,and ; Statistical analysis of pest and disease threat factors in each leaf region and corresponding pest and disease areas. , .

8. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 6, characterized in that, The analysis of the leaf risk area index for identifying different regions within each leaf area includes: Q1. Obtain the leaf surface filamentous texture of each interference area in the dynamic environment of each leaf area in the crop plant image. Combine the effective information of each disease and pest area and identification area in each leaf area in the crop plant image. Construct the contour information of each interference area in the dynamic environment of each leaf area through geometric shape fitting. Integrate it into the complete contour of the leaf vein distribution of each leaf area in the crop plant image. Q2. Identify the complete outline of the leaf vein distribution in each leaf region, the thickness of the corresponding branch veins and the proportion of texture interruption, mark the sub-regions adjacent to the corresponding branch veins in each leaf region, and extract their region color values. Q3. Obtain the standard vein thickness and standard healthy color value of healthy crop leaves, and analyze the stability of the corresponding vein structure in each leaf area accordingly. Q4. Construct a grid to divide the identification area within each leaf region into sub-intervals, denoted as each sub-interval of the identification area within each leaf region. Then, identify the leaf color value and leaf thickness of each sub-interval of the identification area within each leaf region, and evaluate the color brightness of the identification area within each leaf region. Q5. Analyze and determine the leaf risk area index of the identification area in each leaf region.

9. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 1, characterized in that, The method for obtaining the canopy coverage is as follows: The total area of ​​each leaf region is integrated, and the total area of ​​the identification area and the total area of ​​the disease and pest area of ​​each leaf region are obtained. The effective area ratio of each leaf region is then compared. The total number of leaf regions in crop plant images is counted, and the size characteristics of crop plant stems are identified to analyze and calculate the canopy coverage of crop plants.

10. The method for identifying and analyzing crop diseases and pests based on image processing according to claim 1, characterized in that, The formula for assessing the crop plant disease and pest index is as follows: .

Images