Polygonatum sibiricum pest and disease damage prediction method and system based on air-ground cooperation

Abstract

The invention discloses a rhizoma polygonati pest and disease damage prediction method and system based on air-ground collaboration, and the method comprises the steps: obtaining first image information and first data information of rhizoma polygonati, and inputting the first image information and the first data information into a preset dictionary neural network model to obtain pest and disease damage information, the preset dictionary neural network model comprises a polygonatum sibiricum pest and disease damage feature library. According to the method, image information and data information are acquired by using aerial, earth surface and underground acquisition systems, and a dictionary neural network is adopted to perform data analysis to obtain the change speed of diseases and insect pests and the risk probability of occurrence of the diseases and insect pests, so that the efficiency of predicting the diseases and insect pests of polygonatum sibiricum can be improved; and the accuracy and comprehensiveness of polygonatum sibiricum pest and disease damage prediction can be improved, and pest and disease damage treatment and prevention can be carried out.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural plant protection, and in particular to a method and system for predicting diseases and insect pests of Polygonatum sibiricum based on air-ground collaboration. Background Art

[0002] Polygonatum sibiricum is a herb with medicinal value and health care functions. In the planting process, in order to ensure the quality and yield of Polygonatum sibiricum, the prevention of diseases and pests plays a critical role. Most of the existing prediction methods for Polygonatum sibiricum diseases and pests still use manual inspection, which has the disadvantages of low efficiency, poor accuracy and incomplete detection.

[0003] However, the existing intelligent plant protection system is not only unable to effectively prevent the diseases and pests characteristic of Polygonatum sibiricum, but also unable to predict the risks of potential diseases and pests that have not yet occurred, making it difficult to ensure the scientific cultivation and healthy growth of Polygonatum sibiricum. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides a method and system for predicting diseases and pests of Polygonatum cyrtonema based on air-ground collaboration. The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0005] A method and system for predicting Polygonatum sibiricum pests and diseases based on air-ground collaboration, comprising:

[0006] Acquire image data: acquire first image information and first data information of polygonatum;

[0007] Network operation: input the first image information and the first data information into a preset dictionary neural network model to obtain pest and disease information; wherein, the preset dictionary neural network model includes a Polygonatum pest and disease feature library, and the pest and disease information includes the changing speed of pests and diseases and the risk probability of the occurrence of pests and diseases; the first image information includes Polygonatum overview image information, Polygonatum stem and leaf image information, Polygonatum root image information and foreign body image information; the first data information includes meteorological data information, soil data information, time data information and geographic location data information.

[0008] In a specific embodiment, inputting the first image information and the first data information into the preset dictionary neural network model to obtain pest information includes:

[0009] Obtaining a registered two-dimensional image matching the polygonatum pest and disease feature library according to the polygonatum stem and leaf image information and the polygonatum root image information;

[0010] Obtain corresponding pest and disease probability information according to the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign body image information and the polygonatum pest and disease feature library;

[0011] Obtaining a change rate of the pests and diseases according to the pest and disease probability information;

[0012] The risk probability of occurrence of pests and diseases is obtained according to the pest and disease probability information, the meteorological information, the soil information, the time information and the geographical location data information.

[0013] In a specific embodiment, obtaining a registered two-dimensional image matching the polygonatum pest and disease feature library based on the polygonatum stem and leaf image information and the polygonatum root image information includes:

[0014] Obtaining an initial stem and leaf two-dimensional image according to the polygonatum stem and leaf image information;

[0015] Obtaining an initial root two-dimensional image according to the polygonatum root image information;

[0016] Obtaining a registered three-dimensional image according to the initial stem and leaf two-dimensional image and the initial root two-dimensional image;

[0017] The registered two-dimensional image is obtained according to the three-dimensional registered image.

[0018] In a specific embodiment, obtaining a registered three-dimensional image according to the initial stem and leaf two-dimensional image and the initial root two-dimensional image includes:

[0019] According to the initial stem and leaf two-dimensional image and the initial root two-dimensional image, a 360-degree splicing neural network is used to obtain a three-dimensional image of the root of polygonatum sibiricum and a three-dimensional image of the stem and leaf of polygonatum sibiricum;

[0020] The registered three-dimensional image is obtained by fusing the three-dimensional image of the root of polygonatum sibiricum and the three-dimensional image of the stem and leaf of polygonatum sibiricum through an occlusion fusion processing network;

[0021] Among them, the 360-degree splicing neural network includes 5 9*9 convolution kernels, 1 Sigmoid activation function, 4 3*3 convolution kernels and 1 Relu activation function connected in sequence; the occlusion fusion processing network includes 4 7*7 convolution kernels, 1 Sigmoid activation function, 4 5*5 convolution kernels, 1 PRelu activation function and 4 7*7 convolution kernels connected in sequence.

[0022] In a specific embodiment, the corresponding pest and disease probability information is obtained according to the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign body image information and the polygonatum pest and disease feature library, including:

[0023] Obtaining data information of each pest and disease according to the registered two-dimensional image, the soil data information, the foreign matter image information and the Polygonatum sibiricum pest and disease feature library;

[0024] According to the data information of each pest and disease and the polygonatum overview image information, corresponding pest and disease probability information is obtained.

[0025] In a specific embodiment, it also includes:

[0026] According to the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign matter image information and the polygonatum pest and disease feature library, new pest and disease identification information is obtained, including:

[0027] Obtaining new pest and disease data information according to the registered two-dimensional image, the soil data information, the foreign matter image information and the Polygonatum sibiricum pest and disease feature library;

[0028] Obtaining new pests and diseases identification information based on the new pests and diseases data information and the polygonatum sibiricum overview image information;

[0029] When it is determined that the new pest determination information is valid, the second image information and the second data information of the new pest Polygonatum sibiricum at a preset time are input into the Polygonatum sibiricum pest feature library;

[0030] Among them, the second image information includes: an overview image of the new pest and disease Polygonatum sibiricum, images of the stems and leaves of the new pest and disease Polygonatum sibiricum, and images of the roots of the new pest and disease Polygonatum sibiricum; the second data information includes: meteorological data information of the new pest and disease at a preset time, soil data information of the new pest and disease, geographical location data information of the new pest and disease, and time data information of the new pest and disease.

[0031] In a specific implementation, obtaining the change rate of pests and diseases according to the pest and disease probability information includes:

[0032] According to each of the pest probability information, a corresponding pest probability image is obtained;

[0033] According to the probability image of each of the pests and diseases, the change speed of each of the pests and diseases is obtained.

[0034] In a specific embodiment, obtaining the risk probability of occurrence of pests and diseases based on the pest and disease probability information, the meteorological information, the soil information, the time information and the geographic location data information includes:

[0035] Obtain risk data for each pest according to the pest probability information, the meteorological information, the soil information, the time information and the geographical location data information;

[0036] According to the risk data of each pest, the corresponding risk probability of the pest is obtained.

[0037] The present invention also provides a polygonatum pest prediction system based on air-ground collaboration, comprising:

[0038] A data acquisition subsystem, used for acquiring first image information and first data information of polygonatum;

[0039] An information processing system, used for inputting the first image information and the first data information of the polygonatum planting into a preset dictionary neural network model to obtain pest and disease information;

[0040] Among them, the preset dictionary neural network model includes a feature library of Polygonatum pests and diseases, and the pest and disease information includes the changing speed of pests and diseases and the risk probability of the occurrence of pests and diseases; the first image information includes Polygonatum overview image information, Polygonatum stem and leaf image information, Polygonatum root image information and foreign body image information; the first data information includes meteorological data information, soil data information, time data information, and geographic location data information.

[0041] In a specific embodiment, the information processing system includes:

[0042] A single-point sampling registration unit is used to obtain a registered two-dimensional image matching the polygonatum pest and disease feature library according to the polygonatum stem and leaf image information;

[0043] A single-point query analysis unit, used to obtain probability information of each pest and disease and new pest and disease identification information based on the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign body image information and the pest and disease characteristic library;

[0044] A global analysis unit is used to obtain the change rate of the corresponding pests and diseases according to the probability information of each pest and disease;

[0045] A risk prediction unit, used to obtain the risk probability of occurrence of each pest and disease according to the probability information of each pest and disease, the meteorological information, the soil information, the time information and the geographical location data information;

[0046] Wherein, the single-point sampling registration unit includes: a deformation subunit, used to obtain an initial stem and leaf two-dimensional image according to the image information of the polygonatum stem and leaf; a fusion subunit, used to obtain an initial root two-dimensional image according to the image information of the polygonatum root, and to obtain a registered three-dimensional image according to the initial stem and leaf two-dimensional image and the initial root two-dimensional image; a two-dimensional image generation subunit, used to obtain the registered two-dimensional image according to the three-dimensional registered image;

[0047] The single-point query analysis unit includes: a pest query unit, which is used to obtain each pest data information according to the registered two-dimensional image, the soil data information, the foreign body image information and the polygonatum pest feature library; a pest interference determination unit, which is used to obtain corresponding pest probability information according to each pest data information and the polygonatum overview image information;

[0048] The global analysis unit includes: a global determination unit, which is used to obtain a corresponding pest probability image according to each pest probability information; a pest prediction unit, which is used to obtain a corresponding pest change speed according to each pest probability image;

[0049] The risk prediction unit includes: the risk prediction unit includes: a single pest and disease risk prediction unit, which is used to obtain the risk data of each pest and disease based on the pest and disease probability information, the meteorological information, the soil information, the time information and the geographic location data information; a fusion judgment risk prediction unit, which is used to obtain the corresponding pest and disease risk probability based on the risk data of each pest and disease.

[0050] Beneficial effects of the present invention:

[0051] The present invention provides a method and system for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration. By using aerial, surface and underground acquisition systems to obtain image information and data information, and using a dictionary neural network to perform data analysis, the changing speed of diseases and pests and the risk probability of the occurrence of diseases and pests are obtained. This can not only improve the efficiency of prediction of diseases and pests of Polygonatum sibiricum, but also improve the accuracy and comprehensiveness of prediction of diseases and pests of Polygonatum sibiricum, and carry out treatment and prevention of diseases and pests.

[0052] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a method and system flow chart of Polygonatum sibiricum disease and pest prediction based on air-ground collaboration provided by an embodiment of the present invention;

[0054] Figure 2 Schematic diagram of a method and system for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration provided by an embodiment of the present invention;

[0055] Figure 3 It is a schematic diagram of a method for predicting diseases and pests of Polygonatum cyrtonema based on air-ground collaboration and a single-point sampling registration unit in the system provided by an embodiment of the present invention;

[0056] Figure 4 It is a schematic diagram of a method for predicting diseases and insect pests of Polygonatum cyrtonema based on air-ground collaboration and a single-point query analysis unit in the system provided by an embodiment of the present invention;

[0057] Figure 5 It is a schematic diagram of a method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration and a global analysis unit in the system provided by an embodiment of the present invention;

[0058] Figure 6 It is a schematic diagram of a Polygonatum sibiricum disease and pest prediction method based on air-ground collaboration and a risk prediction unit in the system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0060] Embodiment 1

[0061] In one specific embodiment, see Figure 1 , Figure 1 This is a method and system flow chart for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration.

[0062] S1: Obtain image information and data information during the planting process of polygonatum, wherein the image information includes polygonatum overview image information, polygonatum stem and leaf image information, polygonatum root image information and foreign body image information; the data information includes meteorological data information, soil data information, time data information and geographic location data information. The specific image information and data information collection methods include:

[0063] An aerial collection and acquisition device is used to acquire information including an overview image of Polygonatum sibiricum and meteorological data information, wherein the aerial collection and acquisition device may be, for example, a camera drone, an overhead camera, a meteorological information sensor, and other devices that can acquire images of Polygonatum sibiricum. It should be noted that, due to limitations on the shooting distance, the resolution of the shooting device is generally difficult to accurately identify and calculate, so this embodiment simultaneously collects surface and underground information for collaborative processing.

[0064] A surface acquisition device is used to acquire image information including stem and leaf image information of Polygonatum sibiricum and image information of foreign matter at each sampling point, wherein the surface acquisition device includes a surface mobile robot and a low-mount camera;

[0065] An underground acquisition device is used to acquire root image information and soil data information of Polygonatum cyrtonema at each sampling point, wherein the underground acquisition device includes a root monitoring system, a multispectral imaging system and a soil information sensor. When collecting data, a multispectral imaging system is usually placed on the ground to photograph the roots of Polygonatum cyrtonema, and a root monitoring system is usually used as a supplement to be placed in the soil to photograph the roots of Polygonatum cyrtonema;

[0066] The time data information and the geographic location data information are obtained by manual setting, wherein the time data information generally includes the solar terms and the key time nodes of the polygonatum sibiricum planting process, and the geographic location data generally includes the global geographical location of the polygonatum sibiricum planting and the terrain and landforms of the polygonatum sibiricum planting.

[0067] The meteorological data information includes temperature information, precipitation information, humidity information, wind speed and direction, air pressure information, sunshine intensity and time, and extreme weather;

[0068] The foreign body image information includes information on insect pests on the surface of polygonatum and information on insect pests on the ground;

[0069] The soil data information includes soil temperature information, humidity information, pH information, nutrient information, organic matter information and compaction information.

[0070] S2: See Figure 2 , Figure 2 The invention discloses a method and system schematic diagram of a Polygonatum sibiricum pest and disease prediction method based on air-ground collaboration, wherein the image information and data information are input into a preset dictionary neural network model to obtain pest and disease information, wherein the preset dictionary neural network model includes a Polygonatum sibiricum pest and disease feature library, and the pest and disease information includes the changing speed of pests and diseases and the risk probability of pests and diseases appearing, and the system of the preset dictionary neural network model includes multiple single-point sampling registration units, multiple single-point query analysis units, a global analysis unit, a new pest and disease processing unit and a risk prediction unit.

[0071] S21: Since there are differences in the shooting angles and shooting positions of the Polygonatum sibiricum stem and leaf image information and the Polygonatum sibiricum root image information obtained at each sampling collection point, in order to align the collected Polygonatum sibiricum stem and leaf images with the complete Polygonatum sibiricum stem and leaf images in the Polygonatum sibiricum disease and pest feature library, and to align the collected Polygonatum sibiricum root images with the complete Polygonatum sibiricum root images in the Polygonatum sibiricum disease and pest feature library, it is necessary to adopt a three-dimensional image stitching method to perform accurate image registration.

[0072] Since the acquisition equipment for the image information of the stem and leaves of the Polygonatum sibiricum and the image information of the root of the Polygonatum sibiricum obtained at each sampling point may be different, the two images obtained have different colors, brightness and clarity. There is also a missing image at the connection between the image of the stem and leaves of the Polygonatum sibiricum and the image of the root of the Polygonatum sibiricum. Therefore, image fusion technology is required to synthesize the two images into a complete Polygonatum sibiricum image that can match the Polygonatum sibiricum disease and pest feature library.

[0073] Based on the above reasons, it is necessary to synthesize the collected image information of the stems and leaves of polygonatum and the image information of the root of polygonatum into a complete polygonatum image. Specifically:

[0074] According to the image information of the stem and leaf of polygonatum sibiricum and the image information of the root of polygonatum sibiricum obtained at each sampling point, a single-point sampling registration unit is used as a system module, and image processing is performed by a neural network method to generate a registered two-dimensional image matching the polygonatum pest feature library, wherein, see Figure 3 , Figure 3 The invention discloses a method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration and a schematic diagram of a single-point sampling and registration unit in the system. Each single-point sampling and registration unit includes a plurality of deformation subunits, a fusion subunit and a two-dimensional image generation subunit.

[0075] S211: In order to solve the deformation problem and the size inconsistency problem of the collected image, the image information of the stem and leaves of Polygonatum sibiricum and the image information of the root of Polygonatum sibiricum are input into the deformation subunit, and sequentially pass through the first convolution layer, the first pooling layer, the first activation layer and the first upsampling to obtain the initial stem and leaf two-dimensional image and the initial root two-dimensional image. All the initial stem and leaf two-dimensional images are distortion-free images of the same size, and all the initial root two-dimensional images are distortion-free images of the same size, and the initial stem and leaf two-dimensional images match the initial root two-dimensional images in size.

[0076] S212: Input the initial stem and leaf two-dimensional image and the initial root two-dimensional image into the fusion subunit to obtain a registered three-dimensional image.

[0077] S2121: To solve the problem of three-dimensional stitching, for stems and leaves and roots, multi-angle image acquisition is required to obtain multiple initial stem and leaf two-dimensional images and initial root two-dimensional images covering the surface of the stems and leaves and roots, and a 360-degree stitching neural network is used to perform three-dimensional image stitching to obtain a three-dimensional image of the root of Polygonatum cyrtonema and a three-dimensional image of the stem and leaf of Polygonatum cyrtonema;

[0078] The 360-degree splicing neural network method sequentially passes through 5 9*9 convolution kernels, 1 Sigmoid activation function, 4 3*3 convolution kernels and 1 Relu activation function.

[0079] S2122: In order to solve the problem of matching the three-dimensional images of stems and leaves with roots and the problem of missing between the three-dimensional images of stems and leaves and roots, the three-dimensional image of the root of polygonatum sibiricum and the three-dimensional image of the stem and leaves of polygonatum sibiricum are fused by an occlusion fusion processing method to obtain a registered three-dimensional image;

[0080] The occlusion fusion processing method sequentially passes through 4 7*7 convolution kernels, 1 Sigmoid activation function, 4 5*5 convolution kernels, 1 PRelu activation function and 4 7*7 convolution kernels.

[0081] S213: In order to solve the matching problem between the captured image and the pathological image of Polygonatum cyrtonema in the Polygonatum cyrtonema disease and pest feature library, it is necessary to convert the three-dimensional image into a two-dimensional image at any angle, input the three-dimensional registration image to the two-dimensional image generation subunit, and obtain the registered two-dimensional image after passing through the third convolution layer. The registered two-dimensional image matches the diseased Polygonatum cyrtonema image in the Polygonatum cyrtonema disease and pest feature library in terms of quantity, size, angle and position; the registered two-dimensional image needs to maintain high precision, and the image features of local details can be clearly seen, wherein the local details include leaf surface details, stem details and root details.

[0082] S22: Based on the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign body image information and the polygonatum pest and disease feature library, a single-point query analysis unit is used as a system module. Figure 4 , Figure 4 The invention discloses a method for predicting pests and diseases of Polygonatum sibiricum based on air-ground collaboration and a schematic diagram of a single-point query analysis unit in the system. Image processing is performed by dictionary matching and neural network methods to calculate corresponding pest and disease probability information, wherein each single-point query analysis unit includes multiple pest and disease query units and a pest and disease interference determination unit; the number of the pest and disease query units is the same as the number of the single-point sampling registration units and corresponds one to one.

[0083] S221: simultaneously input the registered two-dimensional image, soil data information, the foreign body image information and the polygonatum pest and disease feature library to multiple parallel pest and disease query units, and sequentially pass through the third fully connected layer, the fourth convolutional layer and the third activation layer to obtain each pest and disease data information and new pest and disease data information;

[0084] S222: inputting data information of each pest and disease and the overview image information of Polygonatum cyrtonema into the pest and disease interference determination unit, and sequentially passing through the fifth convolutional layer, the fourth activation layer, the fourth fully connected layer and the fifth activation layer to obtain probability information of each pest and disease;

[0085] S223: inputting new pest data information and polygonatum overview image information to the pest interference determination unit, and sequentially passing through the fifth convolutional layer, the fourth activation layer, the fourth fully connected layer and the fifth activation layer to obtain new pest determination information;

[0086] S23: Based on the pest probability information, a global analysis unit is used as a system module to perform probability analysis through global filtering and neural network methods to obtain the change rate of pests and diseases, wherein, see Figure 5 , Figure 5 The invention discloses a method for predicting diseases and insect pests of Polygonatum sibiricum based on air-ground collaboration and a schematic diagram of a global analysis unit in the system, wherein the global analysis unit comprises a global determination unit and a disease and insect pest prediction unit.

[0087] S231: inputting probability information of each pest into the global determination unit, firstly forming a probability image by sampling the probability of each pest, then corroding and dilating the probability image by using the global filtering method, and finally increasing the accuracy of the probability image by using a two-dimensional interpolation method so as to query the probability image of each pest;

[0088] S232: input each pest probability image to the pest prediction unit, and predict the change rate of each pest through the fifth fully connected layer, the sixth convolutional layer, the sixth fully connected layer and the sixth activation layer in sequence;

[0089] S24: Based on the pest probability information, the meteorological information, the soil information, the time information and the geographical location data information, a risk prediction unit is used as a system module to perform data analysis through a neural network method to obtain the risk probability of pest occurrence, wherein, see Figure 6 , Figure 6 The invention discloses a method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration and a schematic diagram of a risk prediction unit in the system. The global analysis unit includes a plurality of single disease and pest risk prediction units and a fusion judgment risk prediction unit.

[0090] S241: Input the pest probability information, the meteorological information, the soil information, the time information and the geographic location data information into a single pest risk prediction unit, and obtain the risk data of each pest after sequentially passing through the seventh activation layer, the second pooling layer, the seventh convolutional layer and the eighth activation layer;

[0091] S242: input the risk data of each pest to the fusion judgment risk prediction unit, and obtain the risk probability of each pest after passing through the seventh fully connected layer, the eighth convolutional layer and the ninth activation layer in sequence;

[0092] S25: Input new pest information to the new pest processing unit, and use manual processing to update the Polygonatum sibiricum pest feature database, wherein the Polygonatum sibiricum pest feature database includes:

[0093] Overview images of Polygonatum sibiricum at different stages of pests and diseases for each Polygonatum sibiricum;

[0094] The overall image of each Polygonatum sibiricum pest and disease at different pest and disease stages including the roots, stems and leaves of Polygonatum sibiricum;

[0095] Partial images of roots, stems, and leaves of Polygonatum cyrtonema at different stages of each Polygonatum pest and disease;

[0096] The probability of each Polygonatum sibiricum disease and pest triggering other diseases and pests in different disease and pest periods;

[0097] The probability of occurrence of each Polygonatum sibiricum disease and pest under different meteorological environments;

[0098] The probability of occurrence of each Polygonatum sibiricum disease and pest in different geographical locations;

[0099] The probability of occurrence of each Polygonatum disease and pest at different time points;

[0100] The probability of occurrence of each Polygonatum sibiricum disease and pest in different soil environments;

[0101] S251: Determine the validity of the information based on the new pests and diseases;

[0102] S252: If the new pest and disease determination information is valid, record the overview image of the new pest and disease Polygonatum sibiricum, the stem and leaf images of the new pest and disease Polygonatum sibiricum, and the root image of the new pest and disease Polygonatum sibiricum at the preset time into the Polygonatum sibiricum pest and disease feature database; record the meteorological data information, soil data information, geographic location data information and time data information of the new pest and disease Polygonatum sibiricum at the preset time into the Polygonatum sibiricum pest and disease feature database.

[0103] The present invention also provides a polygonatum pest prediction system based on air-ground collaboration, comprising:

[0104] Data acquisition module: obtains the two-dimensional image of the mortise and tenon components to be spliced;

[0105] Neural network processing module: input the two-dimensional image of the mortise and tenon components to be spliced ​​into a preset neural network model to obtain a spliced ​​three-dimensional image or a two-dimensional image of any angle in the spliced ​​three-dimensional image.

[0106] A data acquisition subsystem, used to obtain image information, data information and a characteristic database of diseases and pests of polygonatum sibiricum;

[0107] An information processing system, used for inputting the image information and data information of the polygonatum planting into a preset dictionary neural network model to obtain pest and disease information;

[0108] The single-point sampling registration unit in the information processing system is used to obtain a registered two-dimensional image matching the polygonatum pest and disease feature library according to the polygonatum stem and leaf image information;

[0109] A single-point query analysis unit in the information processing system is used to obtain probability information of each pest and disease and new pest and disease identification information based on the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign body image information and the pest and disease characteristic library;

[0110] The global analysis unit in the information processing system is used to obtain the corresponding change speed of each pest based on the probability information of each pest;

[0111] A risk prediction unit in the information processing system, used to obtain the risk probability of occurrence of each pest and disease based on the pest and disease probability information, the meteorological information, the soil information, the time information and the geographical location data information;

[0112] A new pest and disease processing unit in the information processing system is used to update the Polygonatum sibiricum pest and disease feature database based on new pest and disease information.

[0113] The present invention provides a method and system for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration. By using aerial, surface and underground acquisition systems to obtain image information and data information, and using a dictionary neural network to perform data analysis, the changing speed of diseases and pests and the risk probability of the occurrence of diseases and pests are obtained. This can not only improve the efficiency of prediction of diseases and pests of Polygonatum sibiricum, but also improve the accuracy and comprehensiveness of prediction of diseases and pests of Polygonatum sibiricum, and carry out treatment and prevention of diseases and pests.

[0114] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0115] Although the present application is described herein in conjunction with various embodiments, in the process of implementing the claimed application, those skilled in the art may understand and implement other variations of the disclosed embodiments by reviewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality of components or steps.

[0116] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A method for predicting diseases and insect pests of Polygonatum sibiricum based on air-ground collaboration, characterized in that: Acquire first image information and first data information of Polygonatum sibiricum; Inputting the first image information and the first data information into a preset dictionary neural network model to obtain pest and disease information; Among them, the preset dictionary neural network model includes a feature library of Polygonatum pests and diseases, and the pest and disease information includes the changing speed of pests and diseases and the risk probability of the occurrence of pests and diseases; the first image information includes Polygonatum overview image information, Polygonatum stem and leaf image information, Polygonatum root image information and foreign body image information; the first data information includes meteorological data information, soil data information, time data information and geographic location data information.

2. The method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration according to claim 1 is characterized in that: The step of inputting the first image information and the first data information into the preset dictionary neural network model to obtain pest information includes: Obtaining a registered two-dimensional image matching the polygonatum pest and disease feature library according to the polygonatum stem and leaf image information and the polygonatum root image information; Obtaining pest and disease probability information according to the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign matter image information and the polygonatum pest and disease feature library; Obtaining a change rate of the pests and diseases according to the pest and disease probability information; The risk probability of occurrence of pests and diseases is obtained according to the pest and disease probability information, the meteorological information, the soil information, the time information and the geographical location data information.

3. The method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration according to claim 2 is characterized in that: The method of obtaining a registered two-dimensional image matching the polygonatum pest and disease feature library according to the polygonatum stem and leaf image information and the polygonatum root image information comprises: Obtaining an initial stem and leaf two-dimensional image according to the polygonatum stem and leaf image information; Obtaining an initial root two-dimensional image according to the polygonatum root image information; Obtaining a registered three-dimensional image according to the initial stem and leaf two-dimensional image and the initial root two-dimensional image; The registered two-dimensional image is obtained according to the three-dimensional registered image.

4. The method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration according to claim 3 is characterized in that: Obtaining a registered three-dimensional image according to the initial stem and leaf two-dimensional image and the initial root two-dimensional image, comprising: According to the initial stem and leaf two-dimensional image and the initial root two-dimensional image, a 360-degree splicing neural network is used to obtain a three-dimensional image of the root of polygonatum sibiricum and a three-dimensional image of the stem and leaf of polygonatum sibiricum; The registered three-dimensional image is obtained by fusing the three-dimensional image of the root of polygonatum sibiricum and the three-dimensional image of the stem and leaf of polygonatum sibiricum through an occlusion fusion processing network; Among them, the 360-degree splicing neural network includes 5 9*9 convolution kernels, 1 Sigmoid activation function, 4 3*3 convolution kernels and 1 Relu activation function connected in sequence; the occlusion fusion processing network includes 4 7*7 convolution kernels, 1 Sigmoid activation function, 4 5*5 convolution kernels, 1 PRelu activation function and 4 7*7 convolution kernels connected in sequence.

5. The method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration according to claim 2 is characterized in that: The method of obtaining pest and disease probability information based on the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign body image information and the polygonatum pest and disease feature library includes: Obtaining data information of each pest and disease according to the registered two-dimensional image, the soil data information, the foreign matter image information and the Polygonatum sibiricum pest and disease feature library; According to the data information of each pest and disease and the polygonatum overview image information, corresponding pest and disease probability information is obtained.

6. The method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration according to claim 2 is characterized in that: Also includes: According to the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign matter image information and the polygonatum pest and disease feature library, new pest and disease identification information is obtained, including: Obtaining new pest and disease data information according to the registered two-dimensional image, the soil data information, the foreign matter image information and the Polygonatum sibiricum pest and disease feature library; Obtaining new pests and diseases identification information based on the new pests and diseases data information and the polygonatum sibiricum overview image information; When it is determined that the new pest determination information is valid, the second image information and the second data information of the new pest Polygonatum cyrtonema under the preset time are input into the Polygonatum cyrtonema pest feature database, Among them, the second image information includes: an overview image of the new pest and disease Polygonatum sibiricum, images of the stems and leaves of the new pest and disease Polygonatum sibiricum, and images of the roots of the new pest and disease Polygonatum sibiricum; the second data information includes: meteorological data information of the new pest and disease at a preset time, soil data information of the new pest and disease, geographical location data information of the new pest and disease, and time data information of the new pest and disease.

7. The method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration according to claim 2 is characterized in that: The step of obtaining the change rate of pests and diseases according to the pest and disease probability information includes: According to each of the pest probability information, a corresponding pest probability image is obtained; According to each of the pests and diseases probability images, the corresponding change speed of the pests and diseases is obtained.

8. The method for predicting diseases and pests of Polygonatum sibiricum based on air-ground collaboration according to claim 2 is characterized in that: The step of obtaining the risk probability of occurrence of pests and diseases according to the pest and disease probability information, the meteorological information, the soil information, the time information and the geographical location data information includes: Obtain risk data for each pest according to the pest probability information, the meteorological information, the soil information, the time information and the geographical location data information; According to the risk data of each pest, the corresponding risk probability of the pest is obtained.

9. A polygonatum pest prediction system based on air-ground collaboration, characterized by: include: A data acquisition subsystem, used for acquiring first image information and first data information of polygonatum; An information processing system, used for inputting the first image information and the first data information of the polygonatum planting into a preset dictionary neural network model to obtain pest and disease information; Among them, the preset dictionary neural network model includes a feature library of Polygonatum pests and diseases, and the pest and disease information includes the changing speed of pests and diseases and the risk probability of the occurrence of pests and diseases; the first image information includes Polygonatum overview image information, Polygonatum stem and leaf image information, Polygonatum root image information and foreign body image information; the first data information includes meteorological data information, soil data information, time data information, and geographic location data information.

10. A polygonatum pest prediction system based on air-ground collaboration, characterized by: The information processing system comprises: A single-point sampling registration unit is used to obtain a registered two-dimensional image matching the polygonatum pest and disease feature library according to the polygonatum stem and leaf image information; A single-point query analysis unit, used to obtain pest probability information and new pest determination information based on the registered two-dimensional image, the polygonatum overview image information, the soil data information, the foreign body image information and the pest characteristic library; A global analysis unit, used for obtaining the change speed of corresponding pests and diseases according to the pest and disease probability information; A risk prediction unit, configured to obtain the risk probability of occurrence of each pest or disease based on the pest or disease probability information, the meteorological information, the soil information, the time information and the geographic location data information; Wherein, the single-point sampling registration unit includes: a deformation subunit, used to obtain an initial stem and leaf two-dimensional image according to the image information of the polygonatum stem and leaf; a fusion subunit, used to obtain an initial root two-dimensional image according to the image information of the polygonatum root, and to obtain a registered three-dimensional image according to the initial stem and leaf two-dimensional image and the initial root two-dimensional image; a two-dimensional image generation subunit, used to obtain the registered two-dimensional image according to the three-dimensional registered image; The single-point query analysis unit includes: a pest query unit, which is used to obtain each pest data information according to the registered two-dimensional image, the soil data information, the foreign body image information and the polygonatum pest feature library; a pest interference determination unit, which is used to obtain the corresponding pest probability information according to each pest data information and the polygonatum overview image information; The global analysis unit includes: a global determination unit, which is used to obtain a corresponding pest probability image according to each pest probability information; a pest prediction unit, which is used to obtain the change speed of the corresponding pest according to each pest probability image; The risk prediction unit includes: a single pest and disease risk prediction unit, which is used to obtain the risk data of each pest and disease based on the pest and disease probability information, the meteorological information, the soil information, the time information and the geographic location data information; and a fusion judgment risk prediction unit, which is used to obtain the corresponding pest and disease risk probability based on the risk data of each pest and disease.

Images