Tea tree anthracnose diffusion prediction method based on remote sensing image and lattice point analysis

Abstract

The invention discloses a tea tree anthracnose diffusion prediction method based on a remote sensing image and lattice point analysis, and the method comprises the following steps: obtaining an unmanned plane multispectral image of a tea tree land parcel, and carrying out the recognition of a soil background region, a healthy tea tree region and an infected region; dividing the tea tree plot into uniform grid spaces, defining the state of each grid point as a healthy grid point or an infected grid point, setting a diffusivity calculation method, and calculating the diffusivity of each grid point; setting a disease diffusion rule, updating the state of each grid point according to the disease diffusion rule, and completing the prediction of anthracnose diffusion. According to the invention, by fusing multi-temporal unmanned aerial vehicle remote sensing monitoring data and continuous time sequence environment data, a space-time coupling model for describing a disease infection mechanism between adjacent grid points is constructed, and the spatial diffusion and evolution trend of diseases is dynamically simulated; the problems that a traditional disease prediction method is high in experience dependence, poor in spatial dynamic simulation capability, low in temporal-spatial resolution and the like are solved.

Description

Technical Field

[0001] This invention relates to the field of agricultural engineering technology, and to a method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis. The method first identifies tea trees and soil background based on UAV multispectral imagery, and identifies the distribution of anthracnose disease. After dividing the current area into uniform grids, it integrates continuous temporal environmental data to construct an anthracnose spread prediction model, thereby realizing the spatiotemporal dynamic spread prediction of anthracnose at the tea garden plot scale. Background Technology

[0002] Tea is one of the world's most important economic crops, and anthracnose is one of the most significant diseases affecting tea yield and quality, causing enormous economic losses. Early and precise control of the disease requires accurate prediction of its spread and trends. However, current anthracnose epidemic prediction and spatial diffusion analysis mainly rely on empirical or statistical models. These models are highly dependent on historical data, have low spatial resolution, and poor universality, making it difficult to accurately simulate and predict the dynamic spread paths, local hotspots, and spatiotemporal evolution of the disease at the tea plantation scale, resulting in limited actual control effectiveness. Furthermore, the occurrence and spread of tea anthracnose are influenced by various environmental factors, such as temperature, humidity, wind force, and wind direction. Traditional models often employ simplified assumptions when considering these environmental factors, neglecting their complex interactions. For example, the influence of wind force and direction on the disease's spread distance and direction is often simplified in traditional models, further contributing to low predictive accuracy.

[0003] Therefore, a new method for predicting the spread of anthrax is urgently needed to achieve precise prevention and control. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by providing a method for predicting the spread of anthracnose in tea plants based on remote sensing imagery and grid analysis. By integrating multi-temporal UAV remote sensing data and continuous temporal environmental factors, a spatiotemporally coupled model describing the infection mechanism between adjacent grid points is constructed, dynamically simulating the spatial spread and evolution trends of the disease. This method overcomes the problems of traditional disease prediction methods, such as strong reliance on experience, poor spatial dynamic simulation capabilities, and low spatiotemporal resolution, providing information technology support for the precise control of anthracnose in tea gardens.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis, comprising the following steps: Step S1: Obtain UAV multispectral images of the tea plantation and extract the soil background area, healthy tea plantation area, and infected area. Step S2: Divide the tea plantation into a uniform grid space. Define the state of each grid point as a healthy grid point or an infected grid point, and set a diffusion rate calculation method to calculate the diffusion rate of each grid point. Step S3: Set disease spread rules: For any infected grid point, traverse its eight neighboring grid points and determine the type of the neighboring grid points: If the neighboring grid point is a healthy grid point, then determine the infection based on the spread rate of the healthy grid point; if the neighboring grid point is an infected grid point, then skip the spread operation of the current neighboring grid point and determine the type of the next neighboring grid point; if the neighboring grid point is the soil background, then perform cross-column spread based on distance attenuation; if the neighboring grid point is the boundary of the tea tree plot, then stop the spread. Step S4: Traverse all infected grid points and update the state of each grid point according to the disease spread rules to complete the prediction of anthrax spread.

[0006] Secondly, the present invention provides a field-scale spread prediction system for anthracnose in tea trees, comprising the following modules: The data acquisition module is used to obtain UAV multispectral images and environmental data of the tea plantation to be predicted, and to calculate relevant parameters; The classification module is used to process UAV multispectral images to obtain soil background areas, healthy tea tree areas, and infected areas; The grid division module is used to divide the tea tree plot to be predicted into a uniform grid space containing multiple grid points; The diffusion prediction module calculates the diffusion rate based on environmental data and relevant parameters, updates the status of grid points in the tea plantation to be predicted using the set disease diffusion rules, and completes the prediction of anthracnose diffusion in the tea plantation.

[0007] Thirdly, the present invention provides an electronic device including a processor and a memory, the memory storing machine-executable instructions executable by the processor, the processor executing the machine-executable instructions to implement the above-described method.

[0008] Fourthly, the present invention provides a machine-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to implement the above-described method.

[0009] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention integrates UAV remote sensing data with continuous time-series meteorological environmental data such as temperature, humidity, and wind field data to formulate disease spread rules and predict the spread of anthracnose in tea plantations. It overcomes the problems of traditional disease prediction models, such as strong reliance on experience, poor spatial dynamic simulation capabilities, and low spatiotemporal resolution, thus improving the accuracy and reliability of disease prediction and providing strong support for the precise control of anthracnose in tea gardens. By accurately predicting the spread of the disease, this invention can help tea garden managers take effective control measures in advance, significantly reducing the impact of the disease on tea yield and quality, thereby reducing economic losses. Attached Figure Description

[0010] Figure 1 This is a flowchart of the present invention.

[0011] Figure 2 The image shows the results of region identification in a UAV multispectral image, where A represents the tea tree area and soil background area, and B represents the infected area.

[0012] Figure 3 This is a schematic diagram of the diffusion rules.

[0013] Figure 4 This is a schematic diagram of the diffusion process.

[0014] Figure 5 This is a schematic diagram illustrating the infection status of an example plot. A represents the initial infection status, B represents the actual infection status at time t1, C represents the predicted infection status at time t1, D represents the actual infection status at time t2, and E represents the predicted infection status at time t2.

[0015] Figure 6 The image shows the residuals of the predicted and actual states of a sample plot of land. A represents the residual at time t1, and B represents the residual at time t2. Blue indicates misjudged points, and red indicates missed points. Detailed Implementation

[0016] The present invention will be further described below.

[0017] like Figure 1 As shown, this invention provides a method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis, comprising the following steps: First, UAV multispectral images of the tea plantation are acquired and preprocessed. Based on the preprocessed UAV multispectral images, the red-edge band reflectance of the UAV images is used ( Tea tree row extraction was performed. The mean reflectance of the red-edge band was statistically analyzed. ) and standard deviation ( ), calculate the classification threshold ( Classify the tea trees and background as shown in equations (1.1) and (1.2).

[0018] (1.1) (1.2) in The coefficients, determined through statistical analysis of preprocessed UAV multispectral imagery data, are selected through experiments or optimization processes to adapt to different tea garden conditions. Those below a threshold... The grid points are divided into soil and background. , will be greater than or equal to the threshold The grid points are divided into tea trees Complete the tea tree row extraction operation (such as...) Figure 2 (As shown in A in the diagram).

[0019] Given that anthracnose is a foliar disease, the lesions it forms in the tea tree canopy are clearly characterized in UAV multispectral imagery. The baseline values ​​for healthy and diseased areas of the tea tree used in this invention are obtained through existing methods, which are based on UAV multispectral imagery monitoring results supplemented by ground survey verification data. For example, this embodiment exemplarily provides a segmentation method: Calculate various vegetation indices that reflect the growth status and stress information of tea trees, including but not limited to the normalized vegetation index (NVI). ), Improved Soil-Regulated Vegetation Index ( (e.g.) Because disease-infected areas show significant differences from healthy areas in high-resolution remote sensing images, to adaptively identify infected areas, calculated vegetation indices are used as feature inputs. A combination of K-means and ground validation points is employed to identify classification maps of healthy and infected tea tree areas (e.g., ...). Figure 2 (As shown in B). Typically, the K value can be set to about twice the number of categories, and then the types can be merged based on the ground points to make the results more refined.

[0020] Simultaneously, based on the date of the drone images, environmental data, including the average daily temperature, is obtained from the meteorological station in the location of the tea plantation for a preset period of time prior to the corresponding date. Daily average relative humidity value Average daily wind speed and wind direction Since the monitored tea garden area is a small plot, it is assumed that the environmental factors are the same in all areas of the tea garden plot.

[0021] Then, the relevant parameters required to construct the disease spread rules are calculated; the relevant parameters include the vegetation index (greenness) of the UAV imagery. ), vegetation coverage ( ) and temperature and humidity scores ( ).in, and It is a value that measures the growth status of vegetation, as shown in equations (2.1) and (2.2).

[0022] (2.1) (2.2) in, and These are the near-infrared and green light reflectance values ​​collected by the drone; NDVI is the normalized difference vegetation index. This represents the mean NDVI value of pixels in the purely bare soil within the monitored tea garden area. This represents the mean NDVI value of pure healthy vegetation pixels within the monitored tea garden area.

[0023] Temperature and humidity score This represents the temperature and humidity suitability score for anthracnose pathogens in the monitored tea garden plots. Its expression is shown in equations (2.3), (2.4), and (2.5).

[0024] (2.3) (2.4) (2.5) and These represent the deviations from the optimal temperature and humidity levels, respectively; specifically, when the temperature deviates from the optimal growth and diffusion temperature of anthrax bacteria by 20-30℃, and the humidity deviates from the optimal growth and diffusion humidity by 80-90%,... and The more it decreases, the greater the deviation. and The lower, Preset duration for statistics The arithmetic mean of the products of the fractional deviations in temperature and humidity within a day.

[0025] To eliminate the influence of dimensions, the input vegetation index data and environmental data are normalized to ensure the numerical stability of subsequent calculations.

[0026] The monitored tea garden plots were then set as a uniform grid space, and the status of each grid point was defined as a healthy tea tree or anthracnose-infected tea tree (i.e., a healthy grid point or an infected grid point).

[0027] In this invention, the established disease spread rules include vegetation items. Habitat suitability Spatial distance item and diffusion rate Specifically, the formulas (3.1), (3.2), (3.3), and (3.4) are given below.

[0028] (3.1) (3.2) (3.3) (3.4) in, and It can reflect the sensitivity of vegetation status to diseases. Among them The higher the elevation, the more vigorous the vegetation growth, but the more susceptible it is to infection. The higher the temperature, the stronger the resistance to disease. Depend on and Composition, in which The study characterized the effects of temperature and humidity on the basic suitability of pathogens. The item indicates that the pathogen preferentially depends on the environment, and then infects vegetation. reflect , Complex nonlinear interactions between them. Adjusting parameters. , , Used for weight adjustment to adapt to different plot conditions. Based on the exponential decay model For attenuation distance, attenuation coefficient Modulated by wind field, Indicates the angle between the direction of diffusion and the wind direction: when with the wind, , Decrease, attenuation weakens. and These are the base distance attenuation factor and the wind field modulation factor, respectively. for and Interactive As an overall adjustment coefficient, it regulates the diffusion intensity.

[0029] Let the central infection grid be Its eight neighboring grid points are .in The diffusion radius is defined as the area around the central infected grid point. Using the core as the reference, iterate through its radius. The system uses eight neighboring grid points. For each healthy neighboring point, infection is determined based on the diffusion probability. If a neighboring point is already infected, the diffusion operation for that point is skipped, and the system continues to traverse the next neighboring grid point until the traversal process ends. If the background is soil, cross-column diffusion is achieved using distance decay. Diffusion stops when the system reaches a plot boundary (e.g., ...). Figure 3 (As shown).

[0030] Specifically, disease spread is achieved through spatiotemporally coupled neighborhood interactions. A discrete time step is set. The state update equation is shown in equation (3.5).

[0031] (3.5) in For infection grid The state at time t Indicates position At time t The set of eight neighboring grid points with radius , including spatial coordinates and infection status. When disease originates from an infection point... Spread to neighboring health grids At that time, target grid point vegetation items Habitat suitability Spatial distance item and diffusion probability As shown in equations (3.6), (3.7), (3.8), and (3.9), respectively.

[0032] (3.6) (3.7) (3.8) (3.9) in, , These represent the greenness and vegetation coverage of the target healthy grid points, respectively. The base distance attenuation coefficient for the target grid points is affected by the current wind field: where This indicates the angle between the current wind direction and the direction of diffusion. This represents the Euclidean distance from the infected grid point to the target healthy grid point.

[0033] Set the initial state to t0, and at each time step Perform a state update if a healthy grid point has a radius of [missing information]. There are infected grid points in the eight neighboring regions, and the current spread rate of healthy grid points is... Infection is determined based on the diffusion rate of the healthy grid point: a grid point within the interval is generated. Uniformly distributed random numbers With diffusion rate Compare. If Less than or equal to If successful, it is marked as a new infected grid point; otherwise, the original state is maintained (e.g., ...). Figure 4 (As shown). For example, if the diffusion rate of the current healthy grid point is calculated... If the value is 0.4, then the current healthy grid point has a 0.4 probability of being marked as a new infected grid point, and a 0.6 probability of remaining in its original state.

[0034] In the above process, a genetic algorithm is used to... , , , , , , The parameters were globally optimized to select the optimal parameter combination. The optimized parameters were then used to run a grid model to simulate the spatiotemporal spread of anthrax.

[0035] To further verify the effectiveness of the method of this invention, three independent plots were selected for verification, and the accuracy index between the predicted infection points and the actual observation points was calculated. To evaluate the model's dynamic prediction capability, the earliest observation time was used as the initial observation phase (t0), and the actual observation points and predicted infection points were compared at the intermediate observation phase (t1) and the final observation phase (t2), respectively. And the difference between the real and simulated Moran index The results were quantitatively evaluated.

[0036] in, Used to comprehensively evaluate the accuracy and recall of infection site locations, reflecting the overall performance of prediction results in terms of location accuracy; Moran's index difference The absolute difference between the Moran index calculated from actual observation points and the Moran index calculated based on predicted infection points is used to assess how well the prediction results reproduce the real situation in terms of spatial distribution patterns. The smaller the difference, the more realistic the predicted spatial distribution characteristics are.

[0037] The output evaluation index results are shown in Table 1.

[0038] Table 1. Comparison of F1 score and ΔI for different plots and time phases

[0039] The verification results show that the F1 of the three plots scoreBoth reached over 80%, indicating that the model can predict disease infection points relatively accurately; meanwhile, the difference between the real and simulated Moran's indices... The smaller value indicates that the spatial clustering of disease infection points predicted by the model is close to the actual situation and can better reflect the spatial diffusion characteristics of the disease.

[0040] Taking plot 1 as an example plot, its infection diagrams are as follows: Figure 5 As shown, A represents the initial infection state, B represents the actual infection state at time t1, C represents the predicted infection state at time t1, D represents the actual infection state at time t2, and E represents the predicted infection state at time t2. The residual plot of the actual infection points versus the predicted results for the example plot is shown below. Figure 6 As shown, A is the residual at time t1 and B is the residual at time t2.

[0041] The above embodiments are not intended to limit the present invention, and the present invention is not limited to the above embodiments. Any embodiment that meets the requirements of the present invention falls within the protection scope of the present invention. Furthermore, it should be understood that after reading the above description of the present invention, those skilled in the art can make various alterations or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims.

Claims

1. A method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis, characterized in that, The method includes the following steps: Step S1: Obtain UAV multispectral images of the tea plantation and extract the soil background area, healthy tea plantation area, and infected area. Step S2: Divide the tea plantation into a uniform grid space. Define the state of each grid point as a healthy grid point or an infected grid point, and set a diffusion rate calculation method to calculate the diffusion rate of each grid point. Step S3: Set disease spread rules: For any infected grid point, traverse its eight neighboring grid points and determine the type of the neighboring grid points: If the neighboring grid point is a healthy grid point, then determine the infection based on the spread rate of the healthy grid point; if the neighboring grid point is an infected grid point, then skip the spread operation of the current neighboring grid point and determine the type of the next neighboring grid point; if the neighboring grid point is the soil background, then perform cross-column spread based on distance attenuation; if the neighboring grid point is the boundary of the tea tree plot, then stop the spread. Step S4: Traverse all infected grid points and update the state of each grid point according to the disease spread rules to complete the prediction of anthrax spread.

2. The method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis according to claim 1, characterized in that, In step S1, the red-edge band reflectance of UAV images is used to extract tea tree rows, and the tea tree area and soil background area are obtained. Then, the vegetation index reflecting the growth status and stress information of tea trees in the tea tree area is calculated. Combined with the ground survey method, the tea tree area is divided into healthy tea tree area and infected area. Simultaneously, based on the date of the UAV multispectral imagery, environmental data for a preset time period prior to the corresponding date is acquired, including the average daily temperature value. Daily average relative humidity value Average daily wind speed and wind direction .

3. The method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis according to claim 2, characterized in that, The diffusion rate mentioned in step S2 is calculated using the following method: Calculate relevant parameters from UAV multispectral imagery of tea plantation plots; these parameters include vegetation index (GNDVI), vegetation cover (FVC), and temperature and humidity scores. Es ; Calculate vegetation items based on the relevant parameters and the environmental data. Habitat suitability Spatial distance item Then the diffusivity is obtained. : ； ； ； ； in, , , All are adjustable parameters. For attenuation distance, The attenuation coefficient is... Indicates the angle between the direction of diffusion and the wind direction. This refers to the average daily wind speed. Based on the distance attenuation factor, The wind field modulation factor. This is the overall adjustment coefficient.

4. The method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis according to claim 3, characterized in that, The relevant parameters were calculated using the following method: ； ； in, and These are the near-infrared reflectance and green light reflectance collected by the drone, respectively. Represents the normalized vegetation index. This represents the mean NDVI value of pixels in the purely bare soil within the monitored tea garden area. This represents the mean NDVI value of pixels representing purely healthy vegetation within the monitored tea garden area. ； ； ； in, and These represent the temperature deviation fraction and the humidity deviation fraction, respectively. This is the average daily temperature value. This refers to the daily average relative humidity value; when the temperature deviates from the optimal growth and diffusion temperature of anthrax bacteria, and the humidity deviates from the optimal growth and diffusion humidity, and The more it decreases, the greater the deviation. and The lower; Preset duration for statistics The arithmetic mean of the products of the fractional deviations in temperature and humidity within a day.

5. The method for predicting the spread of anthracnose in tea trees based on remote sensing imagery and grid analysis according to claim 1, characterized in that, In step S3, the spread of the disease is achieved through spatiotemporally coupled neighborhood interactions, with a set discrete time step. The state update equation is shown below: ； in For infection grid The state at time t Indicates infection grid points At time t The set of eight neighborhood grid points with radius , including spatial coordinates and infection status; Set the initial state to t0, and at each time step Perform a state update if a healthy grid point has a radius of [missing information]. There are infected grid points in the eight neighboring regions, and the current spread rate of healthy grid points is... The probability that a currently healthy grid point will be marked as a newly infected grid point is: The probability of maintaining the original state is 1- .

6. A field-scale spread prediction system for tea anthracnose using the method described in any one of claims 1-5, characterized in that, Includes the following modules: The data acquisition module is used to obtain UAV multispectral images and environmental data of the tea plantation to be predicted, and to calculate relevant parameters; The classification module is used to process UAV multispectral images to obtain soil background areas, healthy tea tree areas, and infected areas; The grid division module is used to divide the tea tree plot to be predicted into a uniform grid space containing multiple grid points; The diffusion prediction module calculates the diffusion rate based on environmental data and relevant parameters, updates the status of grid points in the tea plantation to be predicted using the set disease diffusion rules, and completes the prediction of anthracnose diffusion in the tea plantation.

7. An electronic device comprising a processor and a memory, characterized in that, The memory stores machine-executable instructions that can be executed by the processor to implement the method as described in any one of claims 1-5.

8. A machine-readable storage medium storing machine-executable instructions, characterized in that, When the machine-executable instructions are invoked and executed by the processor, the machine-executable instructions cause the processor to implement the method as described in any one of claims 1-5.

Images