Farmland insect pest situation early warning method based on multi-source data fusion

Abstract

The invention discloses a farmland insect situation early warning method based on multi-source data fusion, and relates to a calculation processing system, and the method comprises the following steps: S01, collecting a vegetation image, a topographic image and a climate data set of a target area; s02, creating a plurality of weather charts in one-to-one correspondence with data in the climate data set; s02, grid masks are created for the meteorological charts, the vegetation-integrated images and the topographic images through ArcGIS software, so that a plurality of processed meteorological grid charts, vegetation-integrated grid charts and topographic grid charts are obtained; and S03, inputting the vegetation image set into a pre-trained YOLOV5 target detection model for identification, and extracting a vegetation image with characteristics sufficient to be identified. According to the invention, comprehensive monitoring of farmland environment parameters and vegetation states is realized in a mode of combining unmanned aerial vehicle remote sensing and ground sensor monitoring, the accuracy and timeliness of early warning of diseases and insect pests are improved, a scientific basis is provided for accurate management of farmland, pesticide use is reduced, and agricultural sustainable development is promoted.

Description

Technical Field

[0001] The present invention relates to a computing and processing system, and more particularly to a farmland pest situation early warning method based on multi-source data fusion. Background Art

[0002] Currently, the monitoring of farmland pests and diseases mainly relies on manual inspections or fixed monitoring stations. These methods have problems such as long inspection cycles, limited monitoring ranges, and poor data real-time performance. To solve these problems, a pest and disease monitoring system using drones has emerged.

[0003] For example, in Chinese Patent with the publication number CN113071696A, an agricultural pest and disease monitoring drone is disclosed, which includes a fuselage body. At the bottom of the fuselage body, a monitoring housing and a camera housing are successively arranged from top to bottom. The monitoring housing and the camera housing are rotatably connected. A detection device is fixed inside the monitoring housing. At the bottom of the detection device, there is a camera extending into the camera housing. At the top of the monitoring housing, there is a mosquito repellent device. The mosquito repellent device includes a power housing, a motor, a cam, an air inlet, an exhaust pipe, a piston plate, a piston rod, and a return spring. The fuselage body also includes support legs fixed to its bottom end. Through the mosquito repellent device of the present invention, the surface of the camera housing can be continuously "blown" to prevent the camera from being blocked by flying insects and affecting the shooting effect. With the cooperation of the support legs and the mosquito repellent device, when the whole device lands, it plays a role in decelerating and further has a buffering effect, avoiding a large impact force generated between the device and the ground when it lands at a high speed.

[0004] Or another Chinese Patent with the publication number: CN113139461A, discloses an agricultural planting wheat leaf pest and disease detection system and its management method, which specifically relates to the technical field of agricultural planting. It includes a sampling module. The output end of the sampling module is provided with a detection module. The output end of the detection module is provided with an analysis module. The connection end of the analysis module is provided with a conveying module. The connection end of the conveying module is provided with a display module. The sampling module includes a drone. The detection module includes a PCR detection instrument, which is used to detect the types and quantities of pathogens in the collected soil samples. By installing a soil sampler and a camera on the drone of the present invention to collect soil samples and take photos of wheat leaves, and using the PCR detection instrument to detect whether there are pathogens in the soil samples, it is possible to judge and analyze the causes of the diseased wheat from two aspects of pathogens and pests under the analysis of the central processor, so as to help the wheat field managers more comprehensively understand the reasons for the wheat to be diseased.

[0005] In the prior art including the above two patents, the above systems often focus on the processing of a single data source, such as only analyzing camera images, lacking the comprehensive application of multi-source data, resulting in limited early warning accuracy and efficiency. Summary of the Invention

[0006] The object of the present invention is to provide a farmland pest situation early warning method based on multi-source data fusion to solve the above problems.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A farmland pest situation early warning method based on multi-source data fusion includes the following steps:

[0009] S01. Collect vegetation images, terrain images, and climate data sets of the target area;

[0010] S02. Create multiple meteorological maps corresponding one by one to the data in the climate data set;

[0011] S03. Create grid masks for the multiple meteorological maps, vegetation images, and terrain images through ArcGIS software to obtain multiple processed meteorological grid maps, vegetation grid maps, and terrain grid maps;

[0012] S04. Input the vegetation images into a pre-trained YOLOV5 target detection model for recognition, and extract vegetation pictures with features sufficient to be recognized;

[0013] S05. Analyze through the patent knowledge base according to the vegetation pictures with features sufficient to be recognized, judge the distribution area of pests and diseases on the vegetation images to obtain vegetation images with marks;

[0014] S06. Superimpose and calculate the obtained vegetation images with marks and the vegetation grid maps to obtain a first superimposed layer;

[0015] S07. Then sequentially superimpose and calculate the terrain grid map and the meteorological grid map with the first superimposed layer to obtain a second superimposed layer;

[0016] S08. Extract the pixel points of the pest and disease coverage areas corresponding to the meteorological features in the second superimposed layer and the pest and disease coverage areas corresponding in the first superimposed layer, store them as first data and second data in the cloud library respectively, and create an analysis model.

[0017] Preferably, it further includes the following steps:

[0018] S09. Import the newly collected climate data set into the analysis model to obtain the pest and disease coverage areas.

[0019] Preferably, in step S01, the climate data set includes soil temperature data, wind force data, and air humidity data obtained by sensors arranged at intervals of 50 m in the target area, and light intensity data collected by drones.

[0020] Preferably, the collected light intensity data includes:

[0021] S11. Based on the target area, a takeoff point for UAV sampling is newly established;

[0022] S12. Multiple UAVs converge from the edge of the target area towards the center to collect light intensity data along the line;

[0023] S14. Extract multiple pieces of the light intensity data along the line to obtain the actual brightness change value at adjacent time points and the edge change value of the light intensity along the line, and then import the two change values into the light meteorological map obtained in step 02 for marking;

[0024] S15. Perform superposition calculation on the topographic image and the light meteorological map, and extract the light attenuation rate under the topographic interference characteristics.

[0025] Preferably, the extraction of features sufficient to be recognized in step S04 includes: leaf shape, plant edge contour, and leaf features, and the leaf features include RGB values and HSV values.

[0026] Preferably, the creation of the patent knowledge base includes:

[0027] S41. Determine the crops planted in the target area throughout the year, and collect the disease types of the crops planted throughout the year for at least five years, as well as the disease sample pictures of the plant signs under the disease types;

[0028] S42. Compare and analyze the disease sample pictures of the corresponding plants with the pictures in the normal state, and extract the pixel colors respectively to obtain a pixel set;

[0029] S43. Store the pixel sets corresponding to the crops planted throughout the year.

[0030] Preferably, the line superposition calculation includes:

[0031] S100. Traverse each pixel point of the vegetation image and the grid map to obtain the pixel value of the current pixel point;

[0032] S102. Use the assigned weights to calculate the weighted average value of the current pixel point:

[0033] Weighted average value = (vegetation image pixel value × vegetation image weight) + (grid image pixel value × grid image weight);

[0034] S103. Set the calculated weighted average value as the pixel value of the current pixel point in the superimposed image.

[0035] Preferably, the collected vegetation images in step S01 need to be denoised and the contrast adjusted.

[0036] In the above technical solution, a farmland pest situation early warning method based on multi-source data fusion provided by the present invention has the following beneficial effects: By integrating vegetation images, terrain images, and detailed climate data sets, accurate early warning of farmland pest and disease situations is achieved. It not only improves the accuracy of pest and disease identification, but also provides timely and comprehensive pest and disease distribution information for farmland managers, which helps to guide precise pesticide application and farmland management, reduce pesticide use, and protect the ecological environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flowchart provided for Embodiment 1 of the present invention;

[0038] Figure 2 It is a flowchart of S01 provided for Embodiment 1 of the present invention

[0039] Figure 3 It is a flowchart of S04 provided for Embodiment 1 of the present invention;

[0040] Figure 4 It is a flowchart of the superposition calculation provided for Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To make the technical solution of the present invention clearer, the following further describes the present invention in detail with reference to the accompanying drawings and specific embodiments.

[0042] Embodiment 1

[0043] As Figure 1 shown, a farmland pest situation early warning method based on multi-source data fusion includes the following steps:

[0044] S01. Collect the vegetation images, terrain images, and climate data sets of the target area;

[0045] S02. Create a plurality of meteorological maps corresponding one by one to the data in the climate data set;

[0046] S03. Create grid masks for the plurality of meteorological maps, vegetation images, and terrain images through ArcGIS software to obtain a plurality of processed meteorological grid maps, vegetation grid maps, and terrain grid maps;

[0047] S04. Input the vegetation images into a pre-trained YOLOV5 target detection model for identification, and extract vegetation pictures with features sufficient to be recognized;

[0048] S05. Analyze the vegetation pictures with features sufficient to be recognized through the patent knowledge base to determine the distribution area of pests and diseases on the set of vegetation images, so as to obtain the set of vegetation images with markings.

[0049] S06. Superimpose and calculate the obtained set of vegetation images with markings and the set of vegetation grid maps to obtain the first superimposed layer.

[0050] S07. Then, sequentially superimpose and calculate the terrain grid map and the meteorological grid map with the first superimposed layer to obtain the second superimposed layer.

[0051] S08. Extract the pixel points of the pest and disease coverage areas corresponding to the meteorological features in the second superimposed layer and the pest and disease coverage areas in the first superimposed layer, store them in the cloud library as the first data and the second data respectively, and create an analysis model.

[0052] S09. Import the newly collected climate data set into the analysis model to obtain the pest and disease coverage areas.

[0053] In the above technology, by collecting the climate data set of the target area (including soil temperature, wind force, air humidity, etc.), creating the corresponding meteorological map, and establishing an initial analysis model. Over time, new climate data will be continuously collected. This dependent claim allows these new data to be imported into the established analysis model to update and optimize the model, making it more capable of reflecting the current farmland environmental conditions. Thus, ensuring that the analysis model is always based on the latest climate data, improving the accuracy and timeliness of early warning.

[0054] Further, as Figure 2 shown, the climate data set in step S01 includes the soil temperature data, wind force data, and air humidity data obtained by sensors arranged at intervals of 50 m in the target area, as well as the light intensity data collected by drones.

[0055] And the collected light intensity data includes:

[0056] S11. Newly build a drone sampling take-off point based on the target area;

[0057] S12. Multiple drones converge from the edge of the target area towards the center to collect the light intensity data along the line;

[0058] S14. Extract the multiple light intensity data along the line to obtain the actual brightness change value at adjacent time points and the light intensity edge change value along the line, and then import the two change values into the light meteorological map obtained in step 02 for marking;

[0059] S15. Perform superposition calculation on the terrain image and the light meteorological map, and extract the light attenuation rate under the terrain interference characteristics.

[0060] In the above technology, the drone flies over the target area along a predetermined path while collecting the light intensity data along the route. These data are used to create a light meteorological map and are superimposed and calculated with the topographic image to extract the light attenuation rate under the influence of topographic interference. This method takes into account the influence of topography on light, thereby improving the accuracy of light data. By superimposing the topographic image and the light meteorological map, the influence of topography on light can be evaluated more accurately. Moreover, the combination of high-precision collection by the drone and topographic correction improves the accuracy of light data.

[0061] Secondly, in combination with Figure 3 It can be seen that in step S04, the features sufficient to be recognized are extracted, including: leaf shape, plant edge contour and leaf features, and the leaf features include RGB values and HSV values.

[0062] The creation of the patent knowledge base includes:

[0063] S41. Determine the crops planted throughout the year in the target area, and collect the disease types of the crops planted throughout the year for at least five years, as well as the disease sample pictures of the plant signs under the disease types;

[0064] S42. Compare and analyze the disease sample pictures of the corresponding plants with the pictures in the normal state, and extract the pixel colors respectively to obtain a pixel set;

[0065] S43. Store the pixel sets of the corresponding crops planted throughout the year.

[0066] The above patent knowledge base is established by collecting and analyzing the disease types of the crops planted throughout the year in the target area and the corresponding disease sample pictures of plant signs. These pictures are compared and analyzed with the pictures in the normal state, the key pixel color features are extracted, and stored as a pixel set. During the process of pest and disease identification, these pixel sets are used to match with the currently collected vegetation images to judge the distribution of pests and diseases.

[0067] Furthermore, in combination with Figure 4 It can be seen that the row superposition calculation in the embodiment includes:

[0068] S100. Traverse each pixel point of the vegetation image and the grid map to obtain the pixel value of the current pixel point;

[0069] S102. Use the assigned weights to calculate the weighted average value of the current pixel point:

[0070] Weighted average value = (vegetation image pixel value × vegetation image weight) + (grid image pixel value × grid image weight);

[0071] S103. Set the calculated weighted average value as the pixel value of the current pixel point in the superimposed image.

[0072] The weighted average overlay method is an image processing technique that combines the information of multiple images and generates a new overlay image by assigning different weights to the pixel values of each image. Obtain the labeled set of vegetation images (hereinafter referred to as "vegetation images") and the set of vegetation grid maps (hereinafter referred to as "grid maps") that need to be overlaid.

[0073] Perform necessary preprocessing on the vegetation images and grid maps, such as denoising, adjusting contrast, color balance, etc., to ensure image quality. If the images have different sizes or resolutions, perform scaling or cropping to make them have the same size and resolution. Assign weights to each pixel point of the vegetation images and grid maps. The weights can be determined according to factors such as image clarity, importance, information content, etc. Usually, the weight is a numerical value between 0 and 1, and the sum of all weights should be equal to 1. And using the weighted average overlay method, the labeled set of vegetation images can be accurately overlaid with the set of vegetation grid maps to generate an overlay image containing rich information.

[0074] It should be noted that the set of vegetation images collected in step S01 needs to be denoised and the contrast adjusted.

[0075] In summary, in the first embodiment, by integrating vegetation images, terrain images, and detailed climate data sets, accurate early warning of farmland pests and diseases is achieved. It not only improves the accuracy of pest and disease identification but also provides timely and comprehensive pest and disease distribution information for farmland managers, which helps to guide precise pesticide application and farmland management, reduce pesticide use, and protect the ecological environment.

[0076] Embodiment 2

[0077] The embodiment of the present invention provides a non-transitory computer-readable storage medium, in which at least one instruction or at least one segment of program is stored, and at least one instruction or at least one segment of program is loaded and executed by a processor to implement the steps of:

[0078] Collect the set of vegetation images, terrain images, and climate data sets of the target area;

[0079] Create multiple meteorological maps corresponding one by one to the data in the climate data set;

[0080] Create a grid mask for the multiple meteorological maps, set of vegetation images, and terrain images through ArcGIS software to obtain multiple processed meteorological grid maps, set of vegetation grid maps, and terrain grid maps;

[0081] Input the set of vegetation images into a pre-trained YOLOV5 object detection model for recognition to extract vegetation pictures with features sufficient to be recognized;

[0082] Analyze the vegetation pictures based on the sufficiently recognizable features through the patent knowledge base to determine the distribution area of pests and diseases on the set of vegetation images, so as to obtain a set of vegetation images with identifiers.

[0083] Overlay and calculate the obtained set of vegetation images with identifiers and the set of vegetation grid maps to obtain a first overlay layer.

[0084] Then, sequentially overlay and calculate the terrain grid map and the meteorological grid map with the first overlay layer to obtain a second overlay layer.

[0085] Extract the pixel points of the pest and disease coverage areas corresponding to the meteorological features in the second overlay layer and the pest and disease coverage areas corresponding in the first overlay layer, and store them as first data and second data in the cloud library respectively, and create an analysis model.

[0086] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to memory, storage, database or other media used in the various embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0087] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the above-mentioned division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0088] Embodiment 3

[0089] An embodiment of the present invention provides an electronic device, including a processor and a memory. At least one instruction or at least one program segment is stored in the memory, and the at least one instruction or the at least one program segment is loaded and executed by the processor to implement the steps:

[0090] Collect the vegetation image, terrain image, and climate data set of the target area;

[0091] Create a plurality of meteorological maps corresponding one by one to the data in the climate data set;

[0092] Create a grid mask for the plurality of meteorological maps, vegetation image, and terrain image through ArcGIS software to obtain a plurality of processed meteorological grid maps, vegetation grid maps, and terrain grid maps;

[0093] Input the vegetation image into a pre-trained YOLOV5 object detection model for recognition, and extract vegetation pictures with features sufficient to be recognized;

[0094] Analyze according to the vegetation pictures with features sufficient to be recognized through the patent knowledge base, and judge the distribution area of pests and diseases on the vegetation image to obtain a vegetation image with markings;

[0095] Overlay and calculate the obtained vegetation image with markings and the vegetation grid map to obtain a first overlay layer;

[0096] Then sequentially overlay and calculate the terrain grid map and the meteorological grid map with the first overlay layer to obtain a second overlay layer;

[0097] Lift the pixel points of the pest and disease coverage area corresponding to the meteorological features in the second overlay layer and the pest and disease coverage area corresponding in the first overlay layer, and store them as first data and second data in the cloud library respectively, and create an analysis model.

[0098] The above embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation to the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims

1. A method for early warning of insect pests in farmland based on multi-source data fusion, characterized in that: The following steps are involved: S01, collect vegetation images, terrain images and climate data sets of the target area; S02, creating multiple meteorological maps corresponding to the data in the climate data set; S03, creating a grid mask for the multiple meteorological maps, the vegetation images and the terrain images by using ArcGIS software to obtain multiple processed meteorological grid maps, the vegetation grid maps and the terrain grid maps; S04, inputting the set of vegetation images into a pre-trained YOLOV5 target detection model for recognition, and extracting vegetation images with features sufficient for recognition; S05, analyzing the vegetation image with features sufficient to be identified through a patent knowledge base, determining the distribution area of ​​the pests and diseases on the vegetation image, so as to obtain a vegetation image with a mark; S06, performing superposition calculation on the obtained vegetation collection image with the mark and the vegetation collection grid map to obtain a first superposition layer; S07, then sequentially superimposing the terrain grid map and the meteorological grid map with the first superimposed layer to obtain a second superimposed layer; S08, picking up the pest and disease coverage area corresponding to the meteorological characteristics in the second overlay layer and the pixel points of the pest and disease coverage area corresponding to the first overlay layer, storing them in the cloud library as first data and second data respectively, and creating an analysis model.

2. The method for early warning of insect pests in farmland based on multi-source data fusion according to claim 1, characterized in that: The following steps are also included: S09, importing the newly collected climate data set into the analysis model to obtain the pest and disease coverage area.

3. The method for early warning of insect pests in farmland based on multi-source data fusion according to claim 1, characterized in that: The climate data set in step S01 includes soil temperature data, wind data and air humidity data acquired by sensors arranged at intervals of 50 m in the target area, and light intensity data collected by a drone.

4. The method for early warning of insect pests in farmland based on multi-source data fusion according to claim 3, characterized in that: The collected light intensity data includes: S11, creating a new drone sampling take-off point based on the target area; S12, a plurality of the drones converge from the edge of the target area toward the center to collect light intensity data along the line; S14, extracting a plurality of illumination intensity data along the line to obtain actual brightness change values ​​at adjacent time points and illumination intensity edge change values ​​along the line, and then importing the two change values ​​into the illumination meteorological map obtained in step 02 for marking; S15, superimposing and calculating the terrain image and the light meteorological map to extract the light attenuation rate under the terrain interference characteristics.

5. The method for early warning of insect pests in farmland based on multi-source data fusion according to claim 1, characterized in that: The features extracted in step S04 that are sufficient for identification include: leaf shape, plant edge contour and leaf features, and the leaf features include RGB values ​​and HSV values.

6. The method for early warning of insect pests in farmland based on multi-source data fusion according to claim 1, characterized in that: The creation of the patent knowledge base includes: S41, determining the perennially planted crops in the target area, and collecting disease types of the perennially planted crops for at least five years, as well as disease sample pictures of plant signs under the disease types; S42, comparing and analyzing the diseased image of the corresponding plant with the normal state image, respectively extracting pixel colors to obtain a pixel set; S43, storing the pixel set corresponding to the perennial crops.

7. The method for early warning of insect pests in farmland based on multi-source data fusion according to claim 1, characterized in that: The row superposition calculation includes: S100, traversing each pixel of the vegetation image and the grid map to obtain a pixel value of the current pixel; S102. Calculate the weighted average of the current pixel using the assigned weights: Weighted average = (vegetation image pixel value × vegetation image weight) + (grid image pixel value × grid image weight); S103: Setting the calculated weighted average value as the pixel value of the current pixel in the superimposed image.

8. The method for early warning of insect pests in farmland based on multi-source data fusion according to claim 1, characterized in that: The set of vegetation images collected in step S01 needs to be denoised and have contrast adjusted.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the farmland insect pest early warning method based on multi-source data fusion as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the farmland insect pest early warning method based on multi-source data fusion as described in any one of claims 1 to 8 are implemented.

Images