A rice leaf rice blast detection method based on monitoring images

Abstract

The application discloses a rice leaf rice blast detection method based on a monitoring image, relates to the field of agricultural planting, and solves the problem of poor detection effect of the existing rice leaf rice blast detection method. The method comprises the following steps: S1, periodically acquiring images of a target farmland to be detected to obtain a plurality of rice leaf image sets, performing rice plant leaf image blast analysis on each rice plant in each rice leaf image set, and obtaining leaf image preliminary analysis data according to the analysis result; S2, marking batch disease and pest plants in a farmland region coordinate system according to the leaf image preliminary analysis data to obtain disease and pest rice image marking data; and S3, performing disease and pest plant stage evolution analysis on the target farmland to be detected according to the disease and pest rice image marking data, and issuing a farmland disease and pest warning according to the analysis result. The rice leaf rice blast detection method is comprehensive and accurate.

Description

Technical Field

[0001] This invention belongs to the field of agricultural planting and relates to image monitoring technology, specifically a method for detecting rice blast disease on rice leaves based on monitoring images. Background Technology

[0002] Existing methods for detecting rice leaf blast have the following specific drawbacks when conducting field blast detection:

[0003] 1. Existing methods for detecting rice blast in rice leaves can only provide early warning of blast in individual rice plants by analyzing images of blast in the leaves of a single plant. They cannot perform comprehensive analysis of rice blast images across a field, resulting in a lack of comprehensiveness in the detection results.

[0004] 2. Existing methods for detecting rice blast disease on rice leaves cannot create a planar coordinate system for the target farmland, making it difficult to collect images of rice at different stages to mark rice plants in the farmland area coordinate system in batches. It is also impossible to analyze the stage evolution of blast plants in the target farmland based on the marking results, and it is impossible to issue farmland blast disease warnings based on the analysis results. These methods suffer from the problems of being limited in their detection methods and having weak application and promotion capabilities.

[0005] Therefore, we propose a method for detecting rice blast disease in rice leaves based on monitoring images. Summary of the Invention

[0006] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a method for detecting rice blast disease in rice leaves based on monitoring images. This invention aims to improve the comprehensiveness and applicability of the method for detecting rice blast disease in rice leaves.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] Step S1: Acquire the target farmland to be detected, periodically acquire images of the target farmland to be detected to obtain multiple rice leaf image sets, perform rice blast disease analysis on the rice plants in each rice leaf image set, and obtain preliminary analysis data of the leaf images based on the analysis results.

[0009] Step S2: Create a planar coordinate system for the target farmland to be detected, obtain the farmland area coordinate system, and mark batches of blast diseased plants in the farmland area coordinate system based on the preliminary analysis data of the leaf images to obtain blast diseased rice image marking data.

[0010] Step S3: Based on the image tagging data of rice blast disease, conduct a phased evolution analysis of blast disease plants in the target farmland to be detected, and issue a farmland blast disease warning based on the analysis results.

[0011] Furthermore, step S1 also includes the following steps:

[0012] Step S11: Obtain rice fields that need to be tested for rice blast disease, and randomly select one target rice field to be tested from the multiple rice fields obtained.

[0013] Step S12: During the rice blast detection process in the target farmland, the current time point is marked as the end time point of the cycle to mark a rice pest and disease detection cycle.

[0014] Step S13: Set up several image acquisition batches within the rice pest and disease detection cycle, and name the acquired image acquisition batches P1 to Pa image acquisition batches in chronological order of acquisition time.

[0015] Step S14: Obtain rice leaf images from the P1 image acquisition batch of the target farmland to be detected to form the P1 rice leaf image set; obtain rice leaf images from the P2 image acquisition batch of the target farmland to be detected to form the P2 rice leaf image set; and so on, obtain rice leaf images from the Pa image acquisition batch of the target farmland to be detected to form the Pa rice leaf image set.

[0016] Step S15: Perform blast disease analysis on the P1 rice leaf image set, obtain the plant blast index corresponding to each rice plant based on the analysis results, and obtain the P1 leaf blast analysis data.

[0017] Step S16: Obtain the leaf blast analysis data corresponding to the P2 rice leaf image set to the Pa rice leaf image set respectively, and obtain the P2 leaf blast analysis data to the Pa leaf blast analysis data;

[0018] Step S17: Define the leaf blast analysis data of P1 to leaf blast analysis data of Pa as preliminary leaf image analysis data.

[0019] Furthermore, step S15 also includes the following steps:

[0020] Step S151: Obtain the rice plants planted in the target farmland area to be detected, and randomly select one rice plant from the multiple rice plants obtained as a sample rice plant.

[0021] Step S152: Acquire rice leaf images corresponding to the sample rice plants to obtain multiple rice leaf images, and select one sample rice leaf image from the multiple acquired rice leaf images.

[0022] Step S153: Perform leaf blast analysis on the sample rice leaf images, and obtain the image blast index corresponding to the sample rice leaf images based on the analysis results;

[0023] Step S154: Obtain the image blast index corresponding to each rice leaf image, and calculate the average of the obtained multiple image blast indices to obtain the plant blast index corresponding to the sample rice plant.

[0024] Step S155: Obtain the plant blast index corresponding to each rice plant to obtain the P1 leaf blast analysis data.

[0025] Furthermore, step S153 also includes the following steps:

[0026] Image recognition algorithms are used to mark the rice leaf coverage areas in sample rice leaf images to obtain the image leaf regions;

[0027] The image leaf region is divided into several leaf pixels, and the RGB color model is used to assign the color R value, color G value and color B value to each leaf pixel.

[0028] Multiple historical images of rice leaves were acquired, and the leaf blast disease pixels and healthy leaf pixels in the historical images of rice leaves were all marked.

[0029] Obtain the color R value range, color G value range, and color B value range corresponding to the leaf blast disease pixel points, and obtain the first R value preset range, the first G value preset range, and the first G value preset range.

[0030] Obtain the color R value range, color G value range, and color B value range corresponding to the pixels of healthy leaves, and obtain the second R value preset range, the second G value preset range, and the second G value preset range.

[0031] Leaf pixels whose color R value, color G value, and color B value are respectively located in the first preset range of R value, the first preset range of G value, and the first preset range of B value are obtained to obtain multiple disease area pixels.

[0032] The deviations of the color R value, color G value, and color B value corresponding to each plague region pixel point from the second preset range of R value, the second preset range of G value, and the second preset range of G value are obtained respectively, so as to obtain the color R value deviation, color G value deviation, and color B value deviation corresponding to each plague region pixel point.

[0033] The average of the color R value deviation, color G value deviation, and color B value deviation corresponding to the same plague region pixel is calculated to obtain multiple color RGB deviations.

[0034] The number of leaf pixels in the leaf region of the image is counted to obtain the first pixel count value, and the number of disease pixels in the leaf region of the image is counted to obtain the second pixel count value.

[0035] The image blast index corresponding to the sample rice leaf image is obtained by calculating the color RGB deviation, the number of first pixels, and the number of second pixels.

[0036] The image blast index corresponding to the sample rice leaf images is calculated using the following formula:

[0037]

[0038] Wherein, Wbz is the image blast index corresponding to the sample rice leaf image, Xsz1 is the number of first pixel points, Xsz2 is the number of second pixel points, and Yxs is the color RGB deviation.

[0039] Furthermore, step S2 also includes the following specific steps:

[0040] Step S21: Create a coordinate system for the target farmland to be detected, and obtain the target farmland planar coordinate system;

[0041] Step S22: Obtain preliminary analysis data of leaf images, and obtain leaf blast analysis data from P1 leaf to Pa leaf blast analysis data based on the preliminary analysis data of leaf images;

[0042] Step S23: Based on the leaf blast disease analysis data of P1, determine the growth status type of each rice plant and mark the judgment results on the target farmland plane coordinate system to obtain the P1 stage farmland marking image.

[0043] Step S24: Classify and mark the rice plants in the leaf blast analysis data from P2 to Pa respectively to obtain the staged farmland marking images from P2 to Pa, and obtain the rice blast image marking data.

[0044] Step S23 further includes the following specific steps:

[0045] Based on the leaf blast disease analysis data of P1, the plant blast disease index corresponding to each rice plant in the P1 image acquisition batch was obtained.

[0046] If the plant blast index is 0, the corresponding rice plant is marked as a first-type rice plant; if the plant blast index is greater than 0, the corresponding rice plant is marked as a second-type blast rice plant.

[0047] In the target farmland planar image, the first type of rice plants and the second type of rice plants are marked in the target farmland planar coordinate system to obtain the P1 stage farmland marking image.

[0048] Furthermore, step S3 also includes the following steps:

[0049] Step S31: Obtain image marker data of rice blast disease, and obtain the P1 stage farmland marker image to the Pa stage farmland marker image based on the image marker data of rice blast disease.

[0050] Step S32: Perform blast disease area analysis on the staged farmland marked image of P1, and obtain the blast disease area ratio of the P1 image based on the analysis results;

[0051] Step S33: Obtain the disease area ratios corresponding to the P2 stage farmland marker image to the Pa stage farmland marker image respectively, and obtain the disease area ratios from the P2 image to the Pa image.

[0052] Step S34: Calculate the difference between the plague area ratio in image P2 and the plague area ratio in image P1, and calculate the ratio of the difference to the plague area ratio in image P1 to obtain the plague area change rate of K1. Calculate the difference between the plague area ratio in image P3 and the plague area ratio in image P2, and calculate the ratio of the difference to the plague area ratio in image P2 to obtain the plague area change rate of K2. Similarly, calculate the difference between the plague area ratio in image Pa and the plague area ratio in image Pa-1, and calculate the ratio of the difference to the plague area ratio in image Pa-1 to obtain the plague area change rate of Kd.

[0053] Step S35: Calculate the average of the area change rates of the K1 plague region and the Kd plague region to obtain the comprehensive change coefficient of the plague region;

[0054] Step S36: If the comprehensive variation coefficient of the blast disease area is greater than 1, then the target farmland to be tested is determined to be a farmland where rice blast disease has spread.

[0055] Step S37: If the comprehensive variation coefficient of the blast disease area is less than or equal to 1, then the target farmland to be tested is determined to be rice blast stable farmland.

[0056] Step S38: Analyze the spread trend of rice blast in farmland and issue a blast spread warning based on the analysis results.

[0057] Furthermore, step S32 also includes the following steps:

[0058] The P1 stage farmland marker image is divided into several F1 type plant squares. The number of F1 type plant squares occupied by the first type of rice plants is obtained to obtain the number value of the first plant squares. The number of F1 type plant squares occupied by the second type of rice plants is obtained to obtain the number value of the second plant squares.

[0059] The disease area ratio in image P1 is obtained by calculating the number of squares of the first plant and the number of squares of the second plant.

[0060] The area ratio of the plague in the P1 image is calculated using the following formula:

[0061]

[0062] Where Smp1 is the area ratio of the diseased area in the P1 image, Zgf1 is the number of squares of the first plant, and Zgf2 is the number of squares of the second plant.

[0063] Furthermore, step S38 also includes the following steps:

[0064] Step S381: Among the staged farmland marker images P1 to Pa, set the staged farmland marker images P1 and P2 as the first image analysis group, set the staged farmland marker images P2 and P3 as the second image analysis group, and so on, set the staged farmland marker images Pa-1 and Pa as the e-th image analysis group;

[0065] Step S382: Perform blast disease plant spread analysis on the first image analysis group, and obtain the first blast disease spatial spread value based on the analysis results;

[0066] Step S383: Obtain the plague spatial diffusion values ​​corresponding to the second image analysis group to the e-th image analysis group respectively, and obtain the second plague spatial diffusion value to the e-th plague spatial diffusion value;

[0067] Step S384: Calculate the average value of the spatial diffusion value of the first disease to the eth disease spatial diffusion value to obtain the spatial diffusion value of the disease in the farmland;

[0068] Step S385: Obtain the preset range of the spatial diffusion value of rice blast. If the spatial diffusion value of rice blast is within the range, it is determined that the rice blast-affected farmland is in a concentrated diffusion state, and a concentrated diffusion warning is issued. If the spatial diffusion value of rice blast is not within the range, it is determined that the rice blast-affected farmland is in a distributed diffusion state, and a distributed diffusion warning is issued.

[0069] Furthermore, step S382 also includes the following steps:

[0070] Step S3821: Mark the F1 type plant squares in the P1 stage farmland marking image in the target farmland plane coordinate system, and merge the adjacent F1 type plant squares in the target farmland plane coordinate system to obtain multiple F2 type plant squares.

[0071] Step S3822: For F1 type plant squares, mark the coordinates corresponding to the center point of the square to obtain multiple P1F1 diseased plant coordinates; for F2 type plant squares, obtain the coordinates of the geometric center corresponding to the F2 type plant square to obtain multiple P1F2 diseased plant coordinates.

[0072] Step S3823: Mark the newly added F1 type plant squares in the P2 stage farmland marking image relative to the P1 stage farmland marking image in the target farmland plane coordinate system, and merge the newly added F1 type plant squares that are adjacent in the target farmland plane coordinate system to obtain multiple newly added F2 type plant squares.

[0073] Step S3824: For the newly added F1 type plant square, mark the coordinates corresponding to the center point of the square to obtain the coordinates of multiple P2F1 diseased plants. For the newly added F2 type plant square, obtain the coordinates of the geometric center corresponding to the newly added F2 type plant square to obtain the coordinates of multiple P2F2 diseased plants.

[0074] Step S3825: Perform regional diffusion distance analysis on the F1 type plant grid, and obtain the diffusion distance of the F1 grid based on the analysis results;

[0075] Step S3826: Perform regional diffusion distance analysis on the F2 type plant grid, and obtain the diffusion distance of the F2 grid based on the analysis results;

[0076] Step S3825 further includes the following steps:

[0077] In the target farmland plane coordinate system, select a sample P1F1 diseased plant coordinate from the multiple obtained P1F1 diseased plant coordinates, obtain the coordinate distance value between the sample P1F1 diseased plant coordinate and each P2F1 diseased plant coordinate, and take the average value of the multiple coordinate distance values ​​to obtain the disease spread distance corresponding to the sample P1F1 diseased plant coordinate.

[0078] Obtain the disease spread distance corresponding to the coordinates of each P1F1 diseased plant, and compare the numerical values ​​of the obtained disease spread distances. Mark the disease spread distance with the largest value as the average spread distance of the F1 square.

[0079] Furthermore, step S3826 also includes the following steps:

[0080] In the target farmland plane coordinate system, select a sample P1F2 diseased plant coordinate from the multiple obtained P1F2 diseased plant coordinates, obtain the coordinate distance value between the sample P1F2 diseased plant coordinate and each P2F2 diseased plant coordinate, and take the average value of the multiple coordinate distance values ​​to obtain the disease spread distance corresponding to the sample P1F2 diseased plant coordinate.

[0081] Obtain the disease spread distance corresponding to the coordinates of each P1F2 diseased plant, and compare the numerical values ​​of the obtained disease spread distances. Mark the disease spread distance with the largest value as the second disease spread distance.

[0082] The number of F1 type plant squares in each newly added F2 type plant square is obtained, resulting in multiple F1 square count values. The average value of the multiple F1 square count values ​​is then calculated to obtain the square equivalent conversion ratio.

[0083] The spatial diffusion value of the first plague was obtained by calculating the diffusion distance of F1 square, the diffusion distance of F2 square, and the square equivalent conversion ratio.

[0084] The spatial diffusion value of the first plague is calculated using the following formula:

[0085]

[0086] Where Ksz1 is the spatial diffusion value of the first plague, Kj1 is the diffusion distance of F1 square, Kj2 is the diffusion distance of F2 square, and Zhb is the square equivalent conversion ratio.

[0087] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0088] 1. This invention improves the comprehensiveness of blast disease detection results by analyzing images of rice leaves at different growth stages and providing early warning of blast disease for individual rice plants, while also conducting comprehensive analysis of all rice blast disease images within the farmland area.

[0089] 2. This invention creates a planar coordinate system for the target farmland to be detected, marks rice plants with blast disease in batches based on rice images at different stages, analyzes the stage-by-stage evolution of blast disease plants in the target farmland based on the marking results, and determines the spread range of blast disease rice and issues farmland blast disease warnings based on the analysis results. This invention can functionally expand the blast disease detection method and improve the application scope of the blast disease detection method in various scenarios. Attached Figure Description

[0090] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0091] Figure 1 This is a diagram illustrating the implementation steps of the present invention;

[0092] Figure 2 This is a schematic diagram of the target farmland plane coordinate system of the present invention;

[0093] Figure 3 This is a schematic diagram of the P1 stage farmland marking image of the present invention. Detailed Implementation

[0094] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0095] Example 1

[0096] Please see Figure 1 This invention provides a technical solution: a method for detecting rice blast disease on rice leaves based on monitoring images, comprising the following specific steps:

[0097] Step S1: Acquire the target farmland to be detected, periodically acquire images of the target farmland to be detected to obtain multiple rice leaf image sets, perform rice blast disease analysis on the rice plants in each rice leaf image set, and obtain preliminary analysis data of the leaf images based on the analysis results.

[0098] Step S1 further includes the following steps:

[0099] The rice fields that need to be tested for rice blast disease are acquired, and one target rice field is randomly selected from the acquired rice fields.

[0100] During the process of detecting rice blast in the target farmland, the current time point is marked as the end time point of the cycle to mark a rice pest and disease detection cycle.

[0101] It should be noted here that:

[0102] In this application, the duration of the rice pest and disease detection cycle is generally set to one week. If the rice planting time in the target farmland to be detected is less than one week, the time point corresponding to the completion of rice planting is marked as the start time point of the cycle. If the rice planting time in the target farmland to be detected is less than one week, the time point corresponding to the week before the current time is marked as the start time point of the cycle.

[0103] Several image acquisition batches were set up within the rice pest and disease detection cycle, and the acquired image acquisition batches were named P1 image acquisition batch to Pa image acquisition batch according to the order of acquisition time.

[0104] It should be noted here that:

[0105] In this application, P is the symbol corresponding to the image acquisition batch, and a is the quantity value corresponding to the image acquisition batch, and a is an integer greater than 0.

[0106] Rice leaf images are acquired from the P1 image acquisition batch of the target farmland to be detected to form the P1 rice leaf image set. Rice leaf images are acquired from the P2 image acquisition batch of the target farmland to be detected to form the P2 rice leaf image set, and so on. Rice leaf images are acquired from the Pa image acquisition batch of the target farmland to be detected to form the Pa rice leaf image set.

[0107] It should be noted here that:

[0108] In this application, the rice leaf image sets P1 to Pa, as referred to herein, all include leaf images corresponding to each rice plant in the target farmland to be detected.

[0109] In this application, the image acquisition method and shooting parameters used for the leaf images are the same. The shooting parameters involved here include, but are not limited to, ambient brightness, shooting focal length, and shooting angle.

[0110] Rice leaf blast disease analysis was performed on the P1 rice leaf image set. Based on the analysis results, the plant blast index corresponding to each rice plant was obtained, and P1 leaf blast analysis data was obtained.

[0111] Specifically as follows:

[0112] The rice plants planted in the target farmland area to be tested are obtained, and one rice plant is randomly selected from the obtained rice plants as a sample rice plant.

[0113] The images of rice leaves corresponding to the sample rice plants were acquired, resulting in multiple rice leaf images. Then, one sample rice leaf image was selected from the acquired multiple rice leaf images.

[0114] Leaf blast disease analysis was performed on sample rice leaf images, and the image blast index corresponding to the sample rice leaf images was obtained based on the analysis results.

[0115] Specifically as follows:

[0116] Image recognition algorithms are used to mark the rice leaf coverage areas in sample rice leaf images to obtain the image leaf regions;

[0117] The image leaf region is divided into several leaf pixels, and the RGB color model is used to assign the color R value, color G value and color B value to each leaf pixel.

[0118] Multiple historical images of rice leaves were acquired, and the leaf blast disease pixels and healthy leaf pixels in the historical images of rice leaves were all marked.

[0119] It should be noted here that:

[0120] In this application, the leaf blast disease pixels referred to herein include the blast disease pixels corresponding to each stage of leaf blast development, and the healthy leaf pixels referred to herein include the leaf pixels corresponding to each growth stage of rice leaves.

[0121] Obtain the color R value range, color G value range, and color B value range corresponding to the leaf blast disease pixel points, and obtain the first R value preset range, the first G value preset range, and the first G value preset range.

[0122] Obtain the color R value range, color G value range, and color B value range corresponding to the pixels of healthy leaves, and obtain the second R value preset range, the second G value preset range, and the second G value preset range.

[0123] Leaf pixels whose color R value, color G value, and color B value are respectively located in the first preset range of R value, the first preset range of G value, and the first preset range of B value are obtained to obtain multiple disease area pixels.

[0124] The deviations of the color R value, color G value, and color B value corresponding to each plague region pixel point from the second preset range of R value, the second preset range of G value, and the second preset range of G value are obtained respectively, so as to obtain the color R value deviation, color G value deviation, and color B value deviation corresponding to each plague region pixel point.

[0125] The average of the color R value deviation, color G value deviation, and color B value deviation corresponding to the same plague region pixel is calculated to obtain multiple color RGB deviations.

[0126] The number of leaf pixels in the leaf region of the image is counted to obtain the first pixel count value, and the number of disease pixels in the leaf region of the image is counted to obtain the second pixel count value.

[0127] The image blast index corresponding to the sample rice leaf image is obtained by calculating the color RGB deviation, the number of first pixels, and the number of second pixels.

[0128] The image blast index corresponding to the sample rice leaf images is calculated using the following formula:

[0129]

[0130] Where Wbz is the image blast index corresponding to the sample rice leaf image, Xsz1 is the number of first pixel points, Xsz2 is the number of second pixel points, and Yxs is the color RGB deviation.

[0131] Repeat the process of obtaining the image blast index corresponding to the sample rice leaf image, obtain the image blast index corresponding to each rice leaf image, and calculate the average of the obtained multiple image blast indices to obtain the plant blast index corresponding to the sample rice plant.

[0132] Repeat the process of obtaining the plant blast index corresponding to the sample rice plants, and obtain the plant blast index corresponding to each rice plant to obtain the P1 leaf blast analysis data.

[0133] Repeat the process of acquiring leaf blast analysis data for P1, and acquire leaf blast analysis data corresponding to the rice leaf image sets from P2 to Pa respectively, to obtain leaf blast analysis data from P2 to Pa.

[0134] The leaf blast analysis data from P1 to Pa leaf blast analysis data are defined as preliminary leaf image analysis data.

[0135] Step S2: Create a planar coordinate system for the target farmland to be detected, obtain the farmland area coordinate system, and mark batches of blast diseased plants in the farmland area coordinate system based on the preliminary analysis data of the leaf images to obtain blast diseased rice image marking data.

[0136] Step S2 further includes the following specific steps:

[0137] A coordinate system is created for the target farmland to be detected, resulting in the target farmland's planar coordinate system.

[0138] Specifically as follows:

[0139] Please see Figure 2 The target farmland to be detected is subjected to regional planar image acquisition to obtain a planar image of the target farmland. The geometric center of the image corresponding to the planar image of the target farmland is obtained to obtain the first image feature point. An arbitrary straight line is drawn through the first image feature point to obtain the first image feature line. A straight line perpendicular to the first image feature line is drawn through the first image feature point to obtain the second image feature line.

[0140] The first image feature point is marked as the origin, the first image feature line is marked as the x-axis, and the second image feature line is marked as the y-axis, thus obtaining the target farmland plane coordinate system;

[0141] Acquire preliminary analysis data of leaf images, and based on the preliminary analysis data of leaf images, acquire leaf blast analysis data from P1 leaf to Pa leaf blast analysis data;

[0142] Based on the P1 leaf blast disease analysis data, the growth status of each rice plant was determined, and the results were marked on the target farmland plane coordinate system to obtain the P1 stage farmland marking image.

[0143] Specifically as follows:

[0144] Based on the leaf blast disease analysis data of P1, the plant blast disease index corresponding to each rice plant in the P1 image acquisition batch was obtained.

[0145] If the plant blast index is 0, the corresponding rice plant is marked as a first-type rice plant; if the plant blast index is greater than 0, the corresponding rice plant is marked as a second-type blast rice plant.

[0146] It should be noted here that:

[0147] In this application, the first type of rice plant referred to herein is specifically a rice plant that is not infected with blast disease, and the second type of rice plant referred to herein is specifically a rice plant that is infected with blast disease.

[0148] In the target farmland planar image, the first type of rice plants and the second type of rice plants are marked in the target farmland planar coordinate system to obtain the P1 stage farmland marking image.

[0149] Repeat the process of acquiring the P1 stage farmland marker image, and classify and mark the rice plants in the P2 leaf blast analysis data to the Pa leaf blast analysis data respectively to obtain the P2 stage farmland marker image to the Pa stage farmland marker image, and obtain the blast rice image marker data.

[0150] Step S3: Based on the image tagging data of rice blast disease, conduct a phased evolution analysis of blast disease plants in the target farmland to be detected, and issue a farmland blast disease warning based on the analysis results;

[0151] Step S3 further includes the following steps:

[0152] Obtain image marker data of rice with blast disease, and obtain P1 stage farmland marker images to Pa stage farmland marker images based on the image marker data of rice with blast disease;

[0153] Analyze the area of ​​diseased regions in the P1 stage farmland marker images, and obtain the diseased area ratio in the P1 images based on the analysis results;

[0154] Specifically as follows:

[0155] The P1 stage farmland marker image is divided into several F1 type plant squares. The number of F1 type plant squares occupied by the first type of rice plants is obtained to obtain the number value of the first plant squares. The number of F1 type plant squares occupied by the second type of rice plants is obtained to obtain the number value of the second plant squares.

[0156] It should be noted here that:

[0157] In this application, the F1 type plant grid contains only one rice plant.

[0158] The disease area ratio in image P1 is obtained by calculating the number of squares of the first plant and the number of squares of the second plant.

[0159] The area ratio of the plague in the P1 image is calculated using the following formula:

[0160]

[0161] Where Smp1 is the area ratio of the diseased area in the P1 image, Zgf1 is the number of squares of the first plant, and Zgf2 is the number of squares of the second plant.

[0162] Repeat the process of obtaining the disease area ratio of image P1, and obtain the disease area ratio of the images corresponding to the stage farmland marker images P2 to Pa respectively, to obtain the disease area ratio of image P2 to image Pa.

[0163] Calculate the difference between the plague area ratio in image P2 and the plague area ratio in image P1, and calculate the ratio of the difference to the plague area ratio in image P1 to obtain the plague area change rate of region K1. Calculate the difference between the plague area ratio in image P3 and the plague area ratio in image P2, and calculate the ratio of the difference to the plague area ratio in image P2 to obtain the plague area change rate of region K2. Similarly, calculate the difference between the plague area ratio in image Pa and the plague area ratio in image Pa-1, and calculate the ratio of the difference to the plague area ratio in image Pa-1 to obtain the plague area change rate of region Kd.

[0164] It should be noted here that:

[0165] In this application, K is the symbol corresponding to the rate of change of the plague area, d is the quantitative value corresponding to the rate of change of the plague area, and k = a-1.

[0166] The average of the area change rates of the K1 plague region and the Kd plague region was calculated to obtain the comprehensive change coefficient of the plague region.

[0167] If the comprehensive variation coefficient of the blast disease area is greater than 1, the target farmland to be tested is determined to be a farmland where rice blast disease has spread.

[0168] If the comprehensive variation coefficient of the blast disease area is less than or equal to 1, the target farmland to be tested is determined to be rice blast-stable farmland.

[0169] It should be noted here that:

[0170] In this application, rice blast-stable farmland refers to farmland where rice blast has occurred but the disease is under control and no longer spreads, while rice blast-spreading farmland refers to farmland where rice blast has occurred and the disease is spreading to surrounding healthy areas.

[0171] Analyze the spread trend of rice blast in farmland and issue early warnings of blast spread based on the analysis results;

[0172] Specifically as follows:

[0173] In the P1 stage farmland marker images to the Pa stage farmland marker images, the P1 stage farmland marker images and the P2 stage farmland marker images are set as the first image analysis group, the P2 stage farmland marker images and the P3 stage farmland marker images are set as the second image analysis group, and so on, with the Pa-1 stage farmland marker images and the Pa stage farmland marker images set as the e-th image analysis group;

[0174] It should be noted here that:

[0175] In this application, e refers to the quantity value corresponding to the image analysis group, and e is an integer greater than 0;

[0176] The spread of the first image analysis group was analyzed to determine the spatial spread value of the first disease.

[0177] Specifically as follows:

[0178] Please see Figure 3 The F1 type plant squares in the P1 stage farmland marking image are marked in the target farmland plane coordinate system, and adjacent F1 type plant squares in the target farmland plane coordinate system are merged to obtain multiple F2 type plant squares.

[0179] For F1 type plant squares, the coordinates corresponding to the center point of the square are marked to obtain the coordinates of multiple P1F1 diseased plants. For F2 type plant squares, the coordinates corresponding to the geometric center of the F2 type plant square are obtained to obtain the coordinates of multiple P1F2 diseased plants.

[0180] The newly added F1 type plant squares in the P2 stage farmland marker image relative to the P1 stage farmland marker image are marked in the target farmland plane coordinate system, and the newly added F1 type plant squares that are adjacent in the target farmland plane coordinate system are merged to obtain multiple newly added F2 type plant squares.

[0181] For newly added F1 type plant squares, the coordinates corresponding to the center point of the square are marked to obtain the coordinates of multiple P2F1 diseased plants. For newly added F2 type plant squares, the coordinates corresponding to the geometric center of the newly added F2 type plant squares are obtained to obtain the coordinates of multiple P2F2 diseased plants.

[0182] Regional diffusion distance analysis was performed on the F1 type plant grid, and the diffusion distance of the F1 grid was obtained based on the analysis results.

[0183] Specifically as follows:

[0184] In the target farmland plane coordinate system, select a sample P1F1 diseased plant coordinate from the multiple obtained P1F1 diseased plant coordinates, obtain the coordinate distance value between the sample P1F1 diseased plant coordinate and each P2F1 diseased plant coordinate, and take the average value of the multiple coordinate distance values ​​to obtain the disease spread distance corresponding to the sample P1F1 diseased plant coordinate.

[0185] Repeat the process of obtaining the disease spread distance corresponding to the coordinates of the P1F1 diseased plant, obtain the disease spread distance corresponding to the coordinates of each P1F1 diseased plant, and compare the obtained disease spread distances of multiple plants. Mark the disease spread distance with the largest value as the average spread distance of the F1 grid.

[0186] Regional diffusion distance analysis was performed on the F2 type plant grid, and the diffusion distance of the F2 grid was obtained based on the analysis results.

[0187] Specifically as follows:

[0188] In the target farmland plane coordinate system, select a sample P1F2 diseased plant coordinate from the multiple obtained P1F2 diseased plant coordinates, obtain the coordinate distance value between the sample P1F2 diseased plant coordinate and each P2F2 diseased plant coordinate, and take the average value of the multiple coordinate distance values ​​to obtain the disease spread distance corresponding to the sample P1F2 diseased plant coordinate.

[0189] Repeat the process of obtaining the disease spread distance corresponding to the coordinates of the P1F2 diseased plant, obtain the disease spread distance corresponding to the coordinates of each P1F2 diseased plant, and compare the values ​​of the obtained disease spread distances. Mark the disease spread distance with the largest value as the second disease spread distance.

[0190] The number of F1 type plant squares in each newly added F2 type plant square is obtained, resulting in multiple F1 square count values. The average value of the multiple F1 square count values ​​is then calculated to obtain the square equivalent conversion ratio.

[0191] The spatial diffusion value of the first plague was obtained by calculating the diffusion distance of F1 square, the diffusion distance of F2 square, and the square equivalent conversion ratio.

[0192] The spatial diffusion value of the first plague is calculated using the following formula:

[0193]

[0194] Where Ksz1 is the spatial diffusion value of the first plague, Kj1 is the diffusion distance of F1 square, Kj2 is the diffusion distance of F2 square, and Zhb is the square equivalent conversion ratio;

[0195] Repeat the process of obtaining the spatial diffusion value of the first plague, and obtain the spatial diffusion values ​​of the plague corresponding to the second image analysis group to the eth image analysis group respectively, to obtain the spatial diffusion values ​​of the second plague to the eth plague.

[0196] The average value of the spatial diffusion value of the first plague to the eth plague spatial diffusion value is calculated to obtain the spatial diffusion value of the plague in farmland;

[0197] The system obtains a preset range of farmland blast spatial diffusion values. If the farmland blast spatial diffusion value is within the farmland blast spatial diffusion value range, it is determined that the farmland where rice blast is spreading is in a concentrated diffusion state, and a concentrated diffusion warning is issued. If the farmland blast spatial diffusion value is not within the farmland blast spatial diffusion value range, it is determined that the farmland where rice blast is spreading is in a distributed diffusion state, and a distributed diffusion warning is issued.

[0198] It should be noted here that:

[0199] Several rice planting monitoring cycles in target farmland under concentrated diffusion were obtained. The spatial diffusion value of farmland blast disease corresponding to each rice planting monitoring cycle was obtained. The farmland blast disease spatial diffusion value with the smallest value was marked as the first spatial diffusion baseline value, and the farmland blast disease spatial diffusion value with the largest value was marked as the second spatial diffusion baseline value. The numerical interval formed by the first spatial diffusion baseline value and the second spatial diffusion baseline value was marked as the preset interval of farmland blast disease spatial diffusion value.

[0200] In this application, if a corresponding calculation formula appears, the above calculation formula is a dimensionless calculation. The weighting coefficient, proportional coefficient and other coefficients in the formula are set to quantify each parameter to obtain a result value. The size of the weighting coefficient and proportional coefficient is only required to not affect the proportional relationship between the parameter and the result value.

[0201] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for detecting rice leaf blast based on monitoring images, characterized in that, The method comprises the following steps: Step S1: obtaining the target farmland to be detected, periodically acquiring images of the target farmland to be detected, obtaining a plurality of rice leaf image sets, and performing leaf image rice blast analysis on each rice plant in each rice leaf image set, and obtaining leaf image preliminary analysis data according to the analysis result; Step S2: creating a plane coordinate system for the target farmland to be detected, obtaining a farmland region coordinate system, and marking the batch of plants with rice blast in the farmland region coordinate system according to the leaf image preliminary analysis data to obtain rice blast image marking data; The step S2 further comprises the following specific steps: Step S21: creating a coordinate system for the target farmland to be detected, and obtaining a target farmland plane coordinate system; Step S22: obtaining leaf image preliminary analysis data, and obtaining P1 leaf blast analysis data to Pa leaf blast analysis data according to the leaf image preliminary analysis data; Step S23: determining the growth state type of each rice plant according to the P1 leaf blast analysis data, and marking the determination result in the target farmland plane coordinate system to obtain a P1 stage farmland marking image; Step S24: dividing and marking the coordinates of the rice plants in the P2 leaf blast analysis data to the Pa leaf blast analysis data to obtain a P2 stage farmland marking image to a Pa stage farmland marking image, and obtaining rice blast image marking data; The step S23 further comprises the following specific steps: According to the P1 leaf blast analysis data, the plant blast index corresponding to each rice plant in the P1 image acquisition batch is obtained according to the P1 leaf blast analysis data; If the plant blast index is 0, the corresponding rice plant is marked as a first type of rice plant, and if the plant blast index is greater than 0, the corresponding rice plant is marked as a second type of rice plant; In the target farmland plane image, the first type of rice plant and the second type of rice plant are marked in the target farmland plane coordinate system to obtain a P1 stage farmland marking image; Step S3: performing rice blast plant stage evolution analysis on the target farmland to be detected according to the rice blast image marking data, and issuing a farmland rice blast warning according to the analysis result; The step S3 further comprises the following steps: Step S31: obtaining rice blast image marking data, and obtaining P1 stage farmland marking image to Pa stage farmland marking image according to the rice blast image marking data; Step S32: performing rice blast area analysis on the P1 stage farmland marking image, and obtaining a P1 image rice blast area ratio according to the analysis result; Step S33: obtaining the image rice blast area ratio corresponding to the P2 stage farmland marking image to the Pa stage farmland marking image to obtain a P2 image rice blast area ratio to a Pa image rice blast area ratio; Step S34: calculate the difference between the P2 image area ratio and the P1 image area ratio, and calculate the ratio of the resulting difference to the P1 image area ratio to obtain the K1 area change rate of the rice blast area, and so on, calculate the difference between the Pa image area ratio and the Pa-1 image area ratio, and calculate the ratio of the resulting difference to the Pa-1 image area ratio to obtain the Kd area change rate of the rice blast area; Step S35: average the K1 area change rate of the rice blast area and the Kd area change rate of the rice blast area to obtain the comprehensive change coefficient of the rice blast area; Step S36: if the comprehensive change coefficient of the rice blast area is greater than 1, it is determined that the target farmland to be detected is a rice blast spreading farmland; Step S37: if the comprehensive change coefficient of the rice blast area is less than or equal to 1, it is determined that the target farmland to be detected is a rice blast stable farmland; Step S38: analyze the spreading trend of the rice blast spreading farmland, and issue a rice blast spreading early warning according to the analysis result; In the step S38, the following steps are further included: Step S381: in the P1 periodic farmland marking image to the Pa periodic farmland marking image, the P1 periodic farmland marking image and the P2 periodic farmland marking image are set as the first image analysis group, and so on, the Pa-1 periodic farmland marking image and the Pa periodic farmland marking image are set as the e image analysis group; Step S382: analyze the rice blast plant spread of the first image analysis group, and obtain the first rice blast spatial spread value according to the analysis result; Step S383: obtain the corresponding rice blast spatial spread value of the second image analysis group to the e image analysis group to obtain the second rice blast spatial spread value to the e rice blast spatial spread value; Step S384: and average the first rice blast spatial spread value to the e rice blast spatial spread value to obtain the farmland rice blast spatial spread value; Step S385: obtain the preset interval of the farmland rice blast spatial spread value, if the farmland rice blast spatial spread value is in the farmland rice blast spatial spread value, it is determined that the rice blast spreading farmland is in a concentrated spreading state, and a concentrated spreading early warning is issued, if the farmland rice blast spatial spread value is not in the farmland rice blast spatial spread value, it is determined that the rice blast spreading farmland is in a distributed spreading state, and a distributed spreading early warning is issued.

2. The method of claim 1, wherein the method comprises: In the step S1, the following steps are further included: Step S11: obtain the rice farmland that needs to be detected for rice blast detection, and randomly select a target farmland to be detected from the obtained multiple rice farmlands; Step S12: in the process of detecting the target farmland for rice blast detection, the time point corresponding to the current time is marked as the end time point of a rice disease and pest detection cycle; Step S13: set the P1 image acquisition batch to the Pa image acquisition batch in the rice disease and pest detection cycle; Step S14: obtain the rice leaf image from the P1 image acquisition batch of the target farmland to be detected to form a P1 rice leaf image set, and so on, obtain the rice leaf image from the Pa image acquisition batch of the target farmland to be detected to form a Pa rice leaf image set; Step S15: rice leaf blight analysis is performed on the P1 rice leaf image set, and a plant blight index corresponding to each rice plant is obtained according to an analysis result, so as to obtain P1 leaf blight analysis data; Step S16: leaf blight analysis data corresponding to the P2 rice leaf image set to the Pa rice leaf image set is obtained, so as to obtain P2 leaf blight analysis data to Pa leaf blight analysis data; Step S17: the P1 leaf blight analysis data to the Pa leaf blight analysis data is defined as leaf image preliminary analysis data.

3. The method of claim 2, wherein the method comprises: In the step S15, the following steps are further included: Step S151: rice plants planted in a target farmland area to be detected are obtained, and one of the rice plants is selected as a sample rice plant; Step S152: a sample rice leaf image is obtained from the sample rice plant, so as to obtain a plurality of rice leaf images, and one of the rice leaf images is selected as a sample rice leaf image; Step S153: leaf blight analysis is performed on the sample rice leaf image, and an image blight index corresponding to the sample rice leaf image is obtained according to an analysis result; Step S154: an image blight index corresponding to each rice leaf image is obtained, and a plurality of image blight indexes are averaged, so as to obtain a plant blight index corresponding to the sample rice plant; Step S155: a plant blight index corresponding to each rice plant is obtained, so as to obtain P1 leaf blight analysis data.

4. The method according to claim 3, wherein the method is characterized by, In the step S153, the following steps are further included: An image recognition algorithm is used to mark a rice leaf coverage area in the sample rice leaf image, so as to obtain an image leaf area; The image leaf area is divided into a plurality of leaf pixel points, and an RGB color model is used to obtain a color R value, a color G value, and a color B value corresponding to each leaf pixel point; A plurality of rice leaf historical images are obtained, and leaf blight pixel points and healthy leaf pixel points in the rice leaf historical images have been marked; Color R value intervals, color G value intervals, and color B value intervals corresponding to the leaf blight pixel points are obtained, so as to obtain a first R value preset interval, a first G value preset interval, and a first B value preset interval; Color R value intervals, color G value intervals, and color B value intervals corresponding to the healthy leaf pixel points are obtained, so as to obtain a second R value preset interval, a second G value preset interval, and a second B value preset interval; Leaf pixel points with the color R value, the color G value, and the color B value in the first R value preset interval, the first G value preset interval, and the first B value preset interval are obtained, so as to obtain a plurality of blight area pixel points; Color R value deviations, color G value deviations, and color B value deviations of each blight area pixel point are obtained, so as to obtain a color R value deviation, a color G value deviation, and a color B value deviation of each blight area pixel point. The color R value deviation, the color G value deviation and the color B value deviation corresponding to the same blight area pixel are subjected to average number calculation to obtain a plurality of color RGB deviations; The number of leaf pixels in the image leaf area is counted to obtain a first pixel number value, and the number of blight pixels in the image leaf area is counted to obtain a second pixel number value; The color RGB deviation, the first pixel number value and the second pixel number value are subjected to calculation to obtain an image blight index corresponding to the sample rice leaf image.

5. The method of claim 1, wherein the method comprises: The step S32 further includes the following steps: The P1 stage farmland marking image is divided into a plurality of F1 type plant squares, the number of F1 type plant squares occupied by the first type of rice plants is obtained to obtain a first plant square number value, and the number of F1 type plant squares occupied by the second type of rice plants is obtained to obtain a second plant square number value; The first plant square number value and the second plant square number value are subjected to calculation to obtain a P1 image blight area ratio.

6. The method of claim 1, wherein the method comprises: The step S382 further includes the following steps: Step S3821: The F1 type plant squares in the P1 stage farmland marking image are marked in the target farmland plane coordinate system, and the F1 type plant squares in an adjacent state in the target farmland plane coordinate system are subjected to square fusion to obtain a plurality of F2 type plant squares; Step S3822: For the F1 type plant square, the coordinates corresponding to the square center point are marked to obtain a plurality of P1F1 disease plant coordinates, and for the F2 type plant square, the coordinates of the geometric center corresponding to the F2 type plant square are obtained to obtain a plurality of P1F2 disease plant coordinates; Step S3823: The newly added F1 type plant squares of the P2 stage farmland marking image relative to the P1 stage farmland marking image are marked in the target farmland plane coordinate system, and the newly added F1 type plant squares in an adjacent state in the target farmland plane coordinate system are subjected to square fusion to obtain a plurality of newly added F2 type plant squares; Step S3824: For the newly added F1 type plant square, the coordinates corresponding to the square center point are marked to obtain a plurality of P2F1 disease plant coordinates, and for the newly added F2 type plant square, the coordinates of the geometric center corresponding to the newly added F2 type plant square are obtained to obtain a plurality of P2F2 disease plant coordinates; Step S3825: The F1 type plant square is subjected to regional diffusion distance analysis, and the F1 square diffusion distance is obtained according to the analysis result; Step S3826: The F2 type plant square is subjected to regional diffusion distance analysis, and the F2 square diffusion distance is obtained according to the analysis result; The step S3825 further includes the following steps: In the target farmland plane coordinate system, a sample P1F1 disease plant coordinate is selected from the obtained plurality of P1F1 disease plant coordinates, the coordinate distance values of the sample P1F1 disease plant coordinate and each P2F1 disease plant coordinate are obtained, and the average value of the obtained plurality of coordinate distance values is taken to obtain the disease plant diffusion distance corresponding to the sample P1F1 disease plant coordinate; The disease strain diffusion distance corresponding to each P1F1 disease strain coordinate is obtained respectively, and the obtained multiple disease strain diffusion distances are compared in value, and the disease strain diffusion distance with the maximum value is marked as the F1 grid average diffusion distance.

7. The method of claim 6, wherein the method comprises: The step S3826 further includes the following steps: In the target farmland plane coordinate system, a sample P1F2 disease strain coordinate is selected from the obtained multiple P1F2 disease strain coordinates, the coordinate distance values of the sample P1F2 disease strain coordinate and each P2F2 disease strain coordinate are obtained respectively, and the average value of the obtained multiple coordinate distance values is obtained to obtain the disease strain diffusion distance corresponding to the sample P1F2 disease strain coordinate; The disease strain diffusion distance corresponding to each P1F2 disease strain coordinate is obtained respectively, and the obtained multiple disease strain diffusion distances are compared in value, and the disease strain diffusion distance with the maximum value is marked as the second disease strain diffusion distance; The number of F1 type plant grids in each newly added F2 type plant grid is obtained to obtain multiple F1 grid number values, and the average value of the obtained multiple F1 grid number values is calculated to obtain the grid equivalent conversion ratio; The F1 grid diffusion distance, the F2 grid diffusion distance and the grid equivalent conversion ratio are calculated to obtain a first blight spatial diffusion value; The first blight spatial diffusion value is calculated, and the specific formula is as follows: ； Wherein, Ksz1 is the first blight spatial diffusion value, Kj1 is the F1 grid diffusion distance, Kj2 is the F2 grid diffusion distance, and Zhb is the grid equivalent conversion ratio.

Images