Rice bacterial blight disease severity prediction method and system based on multiple phenotypic parameters

Abstract

The invention relates to a method and system for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters, and relates to the technical field of rice bacterial blight disease degree prediction. The method comprises firstly calculating the rice spectral reflectance corresponding to each plot in a study area based on a multispectral image, and screening characteristic variables based on the chlorophyll content of rice leaves and the water content of rice plants under the stress of rice bacterial blight, and establishing a new spectral index; secondly predicting the chlorophyll content of rice leaves and the water content of rice plants in the study area based on the rice spectral reflectance and a regression model, and then predicting the degree of rice bacterial blight; and then obtaining a rapid indication of the degree of rice bacterial blight disease based on the new spectral index and the prediction, and is suitable for high-throughput rice disease phenotypic monitoring research.

Description

Technical Field

[0001] The invention relates to the technical field of rice bacterial blight disease extent prediction, and in particular to a rice bacterial blight disease extent prediction method and system based on multiple phenotypic parameters. Background Art

[0002] Rice is one of the main food crops, and disease infection is an important factor affecting rice yield. Timely discovery and understanding of the incidence of rice diseases is of great significance for plant protection operations and field management. Traditional disease incidence detection methods are mostly manual, which will be affected by human subjective judgment and environmental changes. In particular, the detection of the incidence of rice in the field requires a lot of manpower, material resources and time. With the rapid development of modern agricultural technology, UAV remote sensing technology provides support for rapid monitoring of field crop phenotypes. UAVs are equipped with RGB cameras, multispectral cameras or hyperspectral cameras to invert crop canopy information, such as chlorophyll content, moisture content, nitrogen content, etc. However, the models established for the inversion of different phenotypic parameters are different, and the data features used are also different. There are few reports on the inversion of multiple phenotypic parameters using the same data features.

[0003] Bacterial leaf blight is one of the three major rice diseases, which seriously affects rice yield and rice quality. Bacterial leaf blight is a bacterial disease caused by infection with Xanthomonas oryzae pv.Oryzae (abbreviated as: Xoo). After Xoo invades through the wounds or stomata of rice, it proliferates in large numbers in the vascular bundle, causing blockage of the vascular bundle, hindering the transportation of nutrients and water in the plant, weakening photosynthesis, reducing leaf pigment content, reducing leaf water content, and causing leaves to turn yellow and wilt. Therefore, there is a correlation between the severity of rice bacterial leaf blight and the chlorophyll content and water content of rice, but few people pay attention to this correlation at present. Summary of the invention

[0004] In response to the above problems, the present invention provides a method and system for predicting the extent of rice bacterial blight disease based on multiple phenotypic parameters, which utilizes multi-spectral remote sensing technology, chlorophyll content and moisture content changes to achieve rapid indication of the extent of rice bacterial blight disease in the field.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters, comprising:

[0007] Acquire a multispectral image of rice in a study area; the study area includes a plurality of plots;

[0008] Based on the multispectral image, calculating the spectral reflectance of rice corresponding to each of the plots;

[0009] The chlorophyll content of the rice leaves corresponding to each of the plots is determined based on the rice spectral reflectance and the first regression model, and the moisture content of the rice plants corresponding to each of the plots is determined based on the rice spectral reflectance and the second regression model; the first regression model is determined based on the first sample data set; the second regression model is determined based on the second sample data set; the first sample data set includes a plurality of sample rice spectral reflectances and the chlorophyll content of the sample rice leaves corresponding to each of the sample rice spectral reflectances, and the second sample data set includes a plurality of sample rice spectral reflectances and the moisture content of the sample rice plants corresponding to each of the sample rice spectral reflectances; the sample rice is rice under bacterial blight stress;

[0010] Determining the incidence of rice bacterial blight corresponding to each of the plots based on the correlation relationship and the chlorophyll content of the rice leaves and the water content of the rice plants corresponding to each of the plots; the correlation relationship is the correlation relationship between the chlorophyll content of the rice leaves, the water content of the rice plants, and the incidence of rice bacterial blight;

[0011] Based on the score value of the variable projection importance index of the first regression model and the score value of the variable projection importance index of the second regression model, the characteristic variables of the rice spectral reflectance are screened to obtain a spectral index;

[0012] The correlation between the spectral index and the degree of rice bacterial blight is calculated, and based on the correlation between the spectral index and the degree of rice bacterial blight, a visual distribution map of the degree of rice bacterial blight disease in the study area is generated.

[0013] Optionally, it also includes:

[0014] Calculating the correlation between the spectral index and the chlorophyll content of the rice leaves, and generating a visualization distribution map of the chlorophyll content of the rice leaves in the study area based on the correlation between the spectral index and the chlorophyll content of the rice leaves;

[0015] and / or, calculating the correlation between the spectral index and the moisture content of the rice plant, and generating a visualization distribution map of the moisture content of the rice plant in the study area based on the correlation between the spectral index and the moisture content of the rice plant;

[0016] And / or, calculating the correlation between the spectral index and the chlorophyll content of the rice leaves, calculating the correlation between the spectral index and the moisture content of the rice plants, and generating a visualization distribution map of multiple phenotypic parameters of rice in the study area based on the correlation between the spectral index and the severity of rice bacterial leaf blight, the correlation between the spectral index and the chlorophyll content of the rice leaves, and the correlation between the spectral index and the moisture content of the rice plants; the multiple phenotypic parameters include the severity of rice bacterial leaf blight, the chlorophyll content of rice leaves, and the moisture content of rice plants.

[0017] Optionally, it also includes: constructing a sample database; the sample database includes a first sample data set, a second sample data set and a third sample data set; the third sample data set includes multiple sample rice bacterial leaf blight incidence levels and the sample rice leaf chlorophyll content and sample rice plant moisture content corresponding to each of the sample rice bacterial leaf blight incidence levels.

[0018] Optionally, the process of determining the chlorophyll content of the sample rice leaves is:

[0019] A SPAD-502 chlorophyll meter was used to determine the chlorophyll content of rice leaves corresponding to each plot in the sample area.

[0020] Optionally, the process of determining the moisture content of the sample rice plants is:

[0021] The moisture content of the sample rice plants corresponding to each plot in the sample area was calculated using the wet basis moisture content method.

[0022] Optionally, the determination process of the first regression model is:

[0023] The first regression model is determined according to a partial least squares regression method and the first sample data set.

[0024] Optionally, the determination process of the second regression model is:

[0025] The second regression model is determined according to a partial least squares regression method and the second sample data set.

[0026] Optionally, the calculating, based on the multispectral image, the spectral reflectance of rice corresponding to each of the cells specifically includes:

[0027] Preprocessing the multispectral image;

[0028] The spectral reflectance of rice is extracted from the preprocessed multispectral image using ENVI software, and then the spectral reflectance of rice corresponding to each plot is determined.

[0029] A rice bacterial blight disease severity prediction system based on multiple phenotypic parameters, comprising:

[0030] A multispectral image acquisition module, used to acquire a multispectral image of rice in a research area; the research area includes a plurality of plots;

[0031] A rice spectral reflectance calculation module, used for calculating the rice spectral reflectance corresponding to each of the cells based on the multispectral image;

[0032] A rice leaf chlorophyll content and rice plant moisture content calculation module, used to determine the rice leaf chlorophyll content corresponding to each of the plots based on the rice spectral reflectance and a first regression model, and to determine the rice plant moisture content corresponding to each of the plots based on the rice spectral reflectance and a second regression model; the first regression model is determined based on a first sample data set; the second regression model is determined based on a second sample data set; the first sample data set includes a plurality of sample rice spectral reflectances and the sample rice leaf chlorophyll content corresponding to each of the sample rice spectral reflectances, and the second sample data set includes a plurality of sample rice spectral reflectances and the sample rice plant moisture content corresponding to each of the sample rice spectral reflectances; the sample rice is rice under bacterial blight stress;

[0033] A rice bacterial blight disease incidence degree determination module is used to determine the rice bacterial blight disease incidence degree corresponding to each of the plots based on the correlation relationship and the rice leaf chlorophyll content and rice plant water content corresponding to each of the plots; the correlation relationship is the correlation relationship between the rice leaf chlorophyll content, the rice plant water content, and the rice bacterial blight disease incidence degree;

[0034] A spectral index determination module, used for screening the characteristic variables of the rice spectral reflectance based on the score value of the variable projection importance index of the first regression model and the score value of the variable projection importance index of the second regression model to obtain a spectral index;

[0035] The module for generating a visualization distribution map of the degree of rice bacterial blight disease is used to calculate the correlation between the spectral index and the degree of rice bacterial blight disease, and based on the correlation between the spectral index and the degree of rice bacterial blight disease, generate a visualization distribution map of the degree of rice bacterial blight disease in the study area.

[0036] Optionally, it also includes:

[0037] A rice leaf chlorophyll content visualization distribution map generation module, used for calculating the correlation between the spectral index and the chlorophyll content of the rice leaves, and generating a rice leaf chlorophyll content visualization distribution map in the study area based on the correlation between the spectral index and the chlorophyll content of the rice leaves;

[0038] and / or, a rice plant moisture content visualization distribution map generating module, used for calculating the correlation between the spectral index and the rice plant moisture content, and generating a rice plant moisture content visualization distribution map in the study area based on the correlation between the spectral index and the rice plant moisture content;

[0039] And / or, a rice multi-phenotypic parameter visualization distribution map generation module is used to calculate the correlation between the spectral index and the chlorophyll content of the rice leaves, calculate the correlation between the spectral index and the moisture content of the rice plant, and generate a visualization distribution map of the multi-phenotypic parameters of rice in the study area based on the correlation between the spectral index and the severity of rice bacterial blight, the correlation between the spectral index and the chlorophyll content of the rice leaves, and the correlation between the spectral index and the moisture content of the rice plant; the multi-phenotypic parameters include the severity of rice bacterial blight, the chlorophyll content of rice leaves and the moisture content of rice plants.

[0040] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0041] (1) Based on the chlorophyll content of rice leaves and the water content of rice plants under rice bacterial leaf blight stress, characteristic variables were screened and new spectral indicators were established; (2) Based on the rice spectral reflectance and regression model, the chlorophyll content of rice leaves and the water content of rice plants in the study area were predicted, and then the severity of rice bacterial leaf blight was predicted; (3) Based on the new spectral indicators and predictions, a rapid indication of the severity of rice bacterial leaf blight was obtained, which is suitable for high-throughput rice disease phenotypic monitoring research. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0043] Figure 1 It is a flow chart of the method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters of the present invention;

[0044] Figure 2It is a schematic diagram of the overall implementation process of the method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters of the present invention;

[0045] Figure 3 A graph showing the relationship between the SPAD value, moisture content and disease severity of rice measured in the field of the present invention;

[0046] Figure 4 This is a Pearson correlation analysis result diagram between the degree of rice bacterial blight disease, SPAD value and moisture content in the field of the present invention;

[0047] Figure 5 This is the result diagram of the PLSR model of the present invention;

[0048] Figure 6 This is a VIP score result diagram of the present invention;

[0049] Figure 7 A scatter plot between the spectral index SI established by the present invention and the degree of rice bacterial blight disease in the field;

[0050] Figure 8 A scatter plot between the spectral index SI established by the present invention and the SPAD value of rice in the field;

[0051] Fig. 9 A scatter plot between the spectral index SI established by the present invention and the moisture content of rice in the field;

[0052] Fig.10 A visualized distribution map of the degree of rice bacterial blight disease in the field of the present invention;

[0053] Fig.11 This is a result diagram of the rice bacterial blight disease severity prediction system based on multiple phenotypic parameters of the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] Embodiment 1

[0057] like Figure 1 As shown, this embodiment provides a method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters, including:

[0058] Step 100: Acquire a multispectral image of rice in a study area; the study area includes a plurality of plots.

[0059] Step 200: Based on the multispectral image, the spectral reflectance of rice corresponding to each of the plots is calculated.

[0060] Step 300: Determine the chlorophyll content of rice leaves corresponding to each of the plots based on the rice spectral reflectance and a first regression model, and determine the moisture content of rice plants corresponding to each of the plots based on the rice spectral reflectance and a second regression model; the first regression model is determined based on a first sample data set; the second regression model is determined based on a second sample data set; the first sample data set includes a plurality of sample rice spectral reflectances and the chlorophyll content of sample rice leaves corresponding to each of the sample rice spectral reflectances, and the second sample data set includes a plurality of sample rice spectral reflectances and the moisture content of sample rice plants corresponding to each of the sample rice spectral reflectances; the sample rice is rice under bacterial blight stress.

[0061] Step 400: Determine the incidence of rice bacterial blight corresponding to each of the plots based on the correlation relationship and the chlorophyll content of rice leaves and the water content of rice plants corresponding to each of the plots; the correlation relationship is the correlation relationship between the chlorophyll content of rice leaves, the water content of rice plants, and the incidence of rice bacterial blight.

[0062] Step 500: Based on the score value of the variable projection importance index (VIP) of the first regression model and the score value of the variable projection importance index of the second regression model, the characteristic variables of the rice spectral reflectance are screened to obtain spectral indicators.

[0063] Step 600: Calculate the correlation between the spectral index and the severity of rice bacterial blight, and based on the correlation between the spectral index and the severity of rice bacterial blight, generate a visualization distribution map of the severity of rice bacterial blight in the study area.

[0064] Furthermore, the method provided in this embodiment also includes:

[0065] The correlation between the spectral index and the chlorophyll content of the rice leaves is calculated, and based on the correlation between the spectral index and the chlorophyll content of the rice leaves, a visualization distribution map of the chlorophyll content of the rice leaves in the study area is generated.

[0066] And / or, calculating the correlation between the spectral index and the moisture content of the rice plant, and generating a visualization distribution map of the moisture content of the rice plant in the study area based on the correlation between the spectral index and the moisture content of the rice plant.

[0067] And / or, calculating the correlation between the spectral index and the chlorophyll content of the rice leaves, calculating the correlation between the spectral index and the moisture content of the rice plants, and generating a visualization distribution map of multiple phenotypic parameters of rice in the study area based on the correlation between the spectral index and the severity of rice bacterial leaf blight, the correlation between the spectral index and the chlorophyll content of the rice leaves, and the correlation between the spectral index and the moisture content of the rice plants; the multiple phenotypic parameters include the severity of rice bacterial leaf blight, the chlorophyll content of rice leaves, and the moisture content of rice plants.

[0068] Furthermore, the method provided in this embodiment also includes: constructing a sample database; the sample database includes a first sample data set, a second sample data set and a third sample data set; the third sample data set includes multiple sample rice bacterial leaf blight incidence levels and the sample rice leaf chlorophyll content and the sample rice plant moisture content corresponding to each of the sample rice bacterial leaf blight incidence levels.

[0069] In the process of constructing the sample database, the process of determining each element is as follows.

[0070] Determination process of spectral reflectance of sample rice:

[0071] A multispectral camera mounted on an unmanned aerial vehicle is used to obtain multispectral images of rice under bacterial blight stress in the sample area. The multispectral images are processed by image stitching and background removal. Then, based on the processed multispectral images, the spectral reflectance of the sample rice in each plot in the sample area is extracted and recorded as X.

[0072] Determination process of chlorophyll content of sample rice leaves:

[0073] Using a SPAD-502 chlorophyll meter, the chlorophyll content of rice leaves corresponding to each plot in the sample area under bacterial blight stress was collected and recorded as Y1.

[0074] An example is: using a SPAD-502 chlorophyll meter to obtain the chlorophyll content of sample rice leaves, and using the SPAD value instead of the leaf chlorophyll content.

[0075] Among them, when measuring, it is necessary to avoid the main leaf veins. Each leaf from the tip to the leaf sheath is regarded as three parts: upper, middle and lower. Three sampling points are randomly selected from each part to measure the SPAD value. The average value of the SPAD values ​​of these 9 sampling points represents the SPAD value of the rice leaf.

[0076] The process of determining the moisture content of sample rice plants:

[0077] The moisture content of the sample rice plants corresponding to each plot in the sample area is calculated using the wet basis moisture content method, which is recorded as Y2. The calculation formula is:

[0078]

[0079] Wherein, WC represents the plant moisture content of the sample rice, FW represents the fresh weight of the sample rice plant (g), and DW represents the dry weight of the sample rice plant (g).

[0080] The process of determining the incidence of rice bacterial blight in the sample:

[0081] The survey on the incidence of bacterial blight of rice in each plot in the sample area was conducted in accordance with the national standard GB / T 17980.19-2000. Three plant protection experts conducted field surveys and scored on the day of the UAV flight operation. The average of the scores of the three experts was used as the final disease severity score, that is, the incidence of bacterial blight of rice in the sample, recorded as Y3.

[0082] On this basis, the correlation relationship is further determined, that is, according to the third data sample set, the correlation between Y1, Y2 and Y3 is calculated, and then the correlation relationship between the chlorophyll content of rice leaves, the moisture content of rice plants, and the incidence of rice bacterial blight is obtained.

[0083] Among them, the correlation was calculated using Pearson correlation analysis, and the correlation coefficient (r) was calculated using SPSS software. r>0 indicates positive correlation, r<0 indicates negative correlation, r=0 indicates no linear relationship, |r|=1 indicates complete linear correlation, 0<|r|<=0.3 indicates extremely low correlation, 0.3<|r|<=0.5 indicates low correlation, 0.5<|r|<=0.8 indicates significant correlation, and |r|>0.8 indicates extremely high correlation.

[0084] The determination process of the first regression model is:

[0085] The first regression model is determined according to a partial least squares regression (PLSR) method and the first sample data set.

[0086] The determination process of the second regression model is:

[0087] The second regression model is determined according to a partial least squares regression (PLSR) method and the second sample data set.

[0088] As a preferred implementation manner, the calculating the spectral reflectance of rice corresponding to each of the cells based on the multispectral image specifically includes:

[0089] The multispectral image is preprocessed; the spectral reflectance of rice is extracted from the preprocessed multispectral image using ENVI software, and then the spectral reflectance of rice corresponding to each of the plots is determined.

[0090] The present embodiment provides a method for predicting the extent of rice bacterial blight disease based on multiple phenotypic parameters, which mainly includes: using the spectral reflectance of rice under bacterial blight stress obtained by a multi-spectral camera carried by an unmanned aerial vehicle to construct a regression model of spectral reflectance and changes in chlorophyll content and water content of rice under bacterial blight stress, selecting corresponding characteristic bands from the model and then constructing a spectral index to evaluate the extent of rice bacterial blight disease, and the spectral index can be used to quickly indicate the field distribution of chlorophyll content of rice leaves, water content of rice plants and the extent of rice bacterial blight. Obviously, the present embodiment can realize the rapid indication of multiple phenotypic parameters such as chlorophyll content of rice leaves, water content of rice plants and the extent of rice bacterial blight under bacterial blight stress.

[0091] Embodiment 2

[0092] First, the method provided in this embodiment is taken as an example to illustrate that a method for predicting the severity of rice bacterial blight disease based on multiple phenotypic parameters is feasible.

[0093] The experimental data or sample data of this embodiment involve two experimental sites, namely the rice experimental base in Longyou County, Quzhou City, Zhejiang Province (29°0′17″N, 119°10′46″E) and the rice experimental base in Zhuji City, Shaoxing City (29°37′22″N, 120°11′40″E). Both locations have a subtropical monsoon climate, which is humid and rainy, suitable for the growth of rice, and both are prevalent areas of rice blight. Rice can naturally develop the disease after planting without additional treatment. There are 60 plots in the experimental area of ​​the Longyou County Rice Experimental Base, and one rice variety is planted in each plot, for a total of 60 varieties. The rice varieties are ASH1 to ASH60, and each plot is 10.6m long and 4.72m wide. The interval between plots is 0.5m. All rice was cultivated in mid-May 2021, manually transplanted to the field on June 26, and harvested in mid-October. There are 13 plots in the experimental area of ​​Zhuji Rice Experimental Base, each plot is 60m long and 5.5m wide. Each plot is planted with a rice variety, namely Yongyou 31, Jiaheyou 2, Yong 1578, Zhongzheyou 8, Chunxian 7860, Yongyou 15, Jiaheyou 7245, Yongyou 7860, Yongyou 1540, Jia 67, Zhejing 100, Zhejing 165 and Huaxiyou 2171. The interval between plots is 0.5m. All rice was cultivated in mid-May 2021, transplanted to the field manually on June 13, and harvested in mid-October. All rice varieties were provided by Zhejiang Academy of Agricultural Sciences, and rice growth management was carried out according to local management methods. Different rice varieties have different resistance to bacterial blight, and the phenotypes in the field are also different.

[0094] like Figure 2 As shown, the method for predicting the degree of rice bacterial blight disease based on changes in chlorophyll and moisture content provided in this embodiment includes the following steps.

[0095] (1) UAV multispectral image acquisition: A UAV equipped with a 25-band multispectral camera was used to collect multispectral images of the experimental base, with a wavelength range of 600-875nm. The multispectral images of the Longyou County rice experimental base were collected on the 16th, 66th and 92nd day after the rice was transplanted into the field, and the multispectral images of the Zhuji City rice experimental base were collected on the 30th, 68th and 106th day after the rice was transplanted into the field, covering the tillering stage, jointing stage, heading stage and filling stage of rice. During the UAV operation, the distance between the UAV and the ground was 25 meters, the flight speed was 2.5m / s, the heading overlap rate was 60%, and the lateral overlap rate was 75%.

[0096] (2) Spectral reflectance (X) extraction: The multispectral images collected by the drone are multiple continuous images with small viewing angles. The photoscan software is used to stitch these images with small viewing angles into an orthophoto of the entire experimental area. The background information is removed by the threshold segmentation method, and the ENVI software is used to select the area of ​​interest for spectral reflectance extraction. The threshold segmentation method is based on the wavelength with the largest difference in reflectance between the background and the rice as the threshold for processing. In this embodiment, the reflectance difference is the largest at a wavelength of 675nm.

[0097] (3) Measurement of rice chlorophyll content (Y1): On the same day of the drone flight operation, the SPAD-502 chlorophyll meter was used to measure the chlorophyll content of rice in each plot in the experimental area, and the SPAD value was used instead of the chlorophyll content. Three rice plants were randomly selected from each plot, and three fully expanded leaves were randomly selected from each rice plant. Each leaf was considered as the upper, middle, and lower parts from the tip to the sheath. Three sampling points were randomly selected from each part to measure the SPAD value. The average value of the SPAD values ​​of these 27 sampling points represented the SPAD value of the rice plant, and then the average value of the three randomly selected rice plants represented the SPAD value of the rice in the plot.

[0098] (4) Measurement of rice moisture content (Y2): On the same day of the UAV flight operation, three rice plants were randomly selected from each plot in the experimental area, and the fresh weight was measured and recorded after being wiped clean. The rice samples were then placed in an oven at 105°C for 30 minutes to kill green, and then set to 70°C to dry to constant weight, and then weighed and recorded for dry weight. The moisture content of rice was calculated using the wet basis moisture content method, and the average of the three samples in each plot represented the moisture content WC of the plot.

[0099] (5) Investigation of the severity of rice bacterial blight in the field (Y3): On the same day as the UAV flight operation, three plant protection experts conducted a field investigation and scored the disease severity in accordance with the national standard GB / T 17980.19-2000. The disease severity scores were 0, 1, 2, 3, 4, and 5, respectively; 0 represents no obvious symptoms, and 1-5 represents increasing disease severity. Finally, the average of the scores of the three experts was used as the final disease severity score.

[0100] (6) In order to establish a universal model, the data from the two experimental areas were combined for comprehensive analysis. Figure 3 (A) shows the distribution of rice SPAD values ​​at different disease severity levels. Figure 3(B) shows the distribution of rice moisture content under different disease severity. Since the experimental area is large, in order to ensure the reliability of the results, when extracting spectral data, each plot in the experimental area of ​​Longyou County Rice Experimental Base was divided into two sub-plots, and each plot in the experimental area of ​​Zhuji City Rice Experimental Base was divided into five sub-plots, so there were 555 samples in total. Pearson correlation analysis was performed on rice SPAD, moisture content and rice disease severity. The results are shown in the figure. Figure 4 As shown, the degree of rice disease is negatively correlated with rice SPAD and moisture content, with correlation coefficients of -0.83 and -0.77, respectively. *p<0.05 indicates that the correlation is significant at 0.05, and all show significant correlation. After the obtained X, Y1, Y2 and Y3 are matched one by one, they are randomly selected in a ratio of 8:2 to form a modeling set and a prediction set. Among them, this step also determines the correlation between the chlorophyll content of rice leaves, the moisture content of rice plants, and the degree of rice bacterial blight, and this correlation is used in the subsequent actual operation process to determine the degree of rice bacterial blight according to the chlorophyll content of rice leaves and the moisture content of rice plants.

[0101] (7) Using X in the modeling set as input and Y1 and Y2 as output, a PLSR model between rice chlorophyll and moisture content and spectral reflectance under rice bacterial blight stress was established to obtain the prediction data corresponding to the modeling set samples. Then, X in the prediction set was brought into the model to obtain the prediction data corresponding to the prediction set samples. PLSR model result phenotype It is feasible to use multispectral data to predict SPAD and moisture content under rice bacterial blight stress. For the prediction set of SPAD, R2 is 0.67 and RMSE is 5.36. Figure 5 (A) As shown; for the prediction set of moisture content, R2 is 0.72 and RMSE is 0.05. Figure 5 (B) As shown. The PLSR model of this step is used in the subsequent actual operation process, and the chlorophyll content of rice leaves and the water content of rice plants can be obtained only based on the spectral reflectance.

[0102] (8) The VIP scores of the PLSR models of rice SPAD and rice moisture content WC were calculated respectively. The results are as follows: Figure 6 As shown in the figure, VIP values ​​greater than 1 are considered important variables. The spectral data were used to predict rice SPAD and rice moisture content, and the results showed that 658nm, 675nm and 698nm were important wavelengths. Based on these three wavelengths, a new spectral index SI was established according to whether it was correlated with the degree of disease.

[0103] The reflectance values ​​of all samples at 658nm, 675nm and 698nm increase with the increase of disease severity. In order to reflect the difference between disease severity, R658+R698 and R658+R675 were calculated respectively, and then the two were added together. The final SI calculation formula is as follows:

[0104] SI = 2*R658+R675+R698;

[0105] Among them, R658, R675 and R698 represent the spectral reflectance values ​​corresponding to wavelengths of 658nm, 675nm and 698nm respectively.

[0106] There is a positive correlation between SI and disease severity, such as Figure 7 As shown in the figure, the correlation coefficient is 0.762, which proves that the spectral index SI for assessing the degree of rice bacterial blight disease based on rice SPAD and moisture content is effective. Using 0.1 as the threshold can quickly distinguish healthy rice from susceptible rice. SI <= 0.1 is considered healthy rice without disease, and SI > 0.1 is considered susceptible rice. In addition, the correlation between SI and rice SPAD and moisture content is shown in Figure 8 and Fig. 9 As shown, they all show a negative correlation and a significant correlation. Based on the SI value of the corresponding disease degree, we can understand the corresponding situation of SPAD and moisture content.

[0107] (9) Based on the established spectral index SI, the degree of rice disease in the field can be visualized and analyzed. It can also reflect the SPAD and moisture content in the field. The visualization distribution results are as follows: Fig.10 As shown, due to limited space, only the second flight images of the experimental area in the Longyou County Rice Experimental Base are selected for display.

[0108] Embodiment 3

[0109] like Fig.11 As shown, this embodiment provides a rice bacterial blight disease severity prediction system based on multiple phenotypic parameters, including:

[0110] The multispectral image acquisition module 1 is used to acquire the multispectral image of rice in the research area; the research area includes a plurality of plots.

[0111] The rice spectral reflectance calculation module 2 is used to calculate the rice spectral reflectance corresponding to each of the cells based on the multispectral image.

[0112] The rice leaf chlorophyll content and rice plant moisture content calculation module 3 is used to determine the rice leaf chlorophyll content corresponding to each of the plots based on the rice spectral reflectance and a first regression model, and to determine the rice plant moisture content corresponding to each of the plots based on the rice spectral reflectance and a second regression model; the first regression model is determined based on a first sample data set; the second regression model is determined based on a second sample data set; the first sample data set includes a plurality of sample rice spectral reflectances and the sample rice leaf chlorophyll content corresponding to each of the sample rice spectral reflectances, and the second sample data set includes a plurality of sample rice spectral reflectances and the sample rice plant moisture content corresponding to each of the sample rice spectral reflectances; the sample rice is rice under bacterial blight stress.

[0113] The rice bacterial blight disease severity determination module 4 is used to determine the rice bacterial blight disease severity corresponding to each of the plots based on the correlation relationship and the rice leaf chlorophyll content and rice plant water content corresponding to each of the plots; the correlation relationship is the correlation relationship between the rice leaf chlorophyll content, the rice plant water content, and the rice bacterial blight disease severity.

[0114] The spectral index determination module 5 is used to screen the characteristic variables of the rice spectral reflectance based on the score value of the variable projection importance index of the first regression model and the score value of the variable projection importance index of the second regression model to obtain the spectral index.

[0115] The rice bacterial blight disease severity visualization distribution map generation module 6 is used to calculate the correlation between the spectral index and the rice bacterial blight disease severity, and based on the correlation between the spectral index and the rice bacterial blight disease severity, generate a rice bacterial blight disease severity visualization distribution map in the study area.

[0116] Furthermore, the system provided in this embodiment also includes:

[0117] The rice leaf chlorophyll content visualization distribution map generation module is used to calculate the correlation between the spectral index and the chlorophyll content of the rice leaves, and based on the correlation between the spectral index and the chlorophyll content of the rice leaves, generate a rice leaf chlorophyll content visualization distribution map in the study area.

[0118] And / or, a rice plant moisture content visualization distribution map generation module is used to calculate the correlation between the spectral index and the rice plant moisture content, and generate a rice plant moisture content visualization distribution map in the study area based on the correlation between the spectral index and the rice plant moisture content.

[0119] And / or, a rice multi-phenotypic parameter visualization distribution map generation module is used to calculate the correlation between the spectral index and the chlorophyll content of the rice leaves, calculate the correlation between the spectral index and the moisture content of the rice plant, and generate a visualization distribution map of the multi-phenotypic parameters of rice in the study area based on the correlation between the spectral index and the severity of rice bacterial blight, the correlation between the spectral index and the chlorophyll content of the rice leaves, and the correlation between the spectral index and the moisture content of the rice plant; the multi-phenotypic parameters include the severity of rice bacterial blight, the chlorophyll content of rice leaves and the moisture content of rice plants.

[0120] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0121] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for predicting the severity of rice bacterial blight based on multiple phenotypic parameters, characterized in that: include: Acquire a multispectral image of rice in a study area; the study area includes a plurality of plots; Based on the multispectral image, calculating the spectral reflectance of rice corresponding to each of the plots; Determine the chlorophyll content of rice leaves corresponding to each of the plots based on the rice spectral reflectance and the first regression model, and determine the moisture content of rice plants corresponding to each of the plots based on the rice spectral reflectance and the second regression model; The first regression model is determined based on the first sample data set; the second regression model is determined based on the second sample data set; the first sample data set includes a plurality of sample rice spectral reflectances and a sample rice leaf chlorophyll content corresponding to each of the sample rice spectral reflectances, and the second sample data set includes a plurality of sample rice spectral reflectances and a sample rice plant moisture content corresponding to each of the sample rice spectral reflectances; the sample rice is rice under bacterial blight stress; Determining the incidence of rice bacterial blight corresponding to each of the plots based on the correlation relationship and the chlorophyll content of the rice leaves and the water content of the rice plants corresponding to each of the plots; the correlation relationship is the correlation relationship between the chlorophyll content of the rice leaves, the water content of the rice plants, and the incidence of rice bacterial blight; Based on the score value of the variable projection importance index of the first regression model and the score value of the variable projection importance index of the second regression model, the characteristic variables of the rice spectral reflectance are screened to obtain a spectral index; The correlation between the spectral index and the degree of rice bacterial blight is calculated, and based on the correlation between the spectral index and the degree of rice bacterial blight, a visual distribution map of the degree of rice bacterial blight disease in the study area is generated.

2. The method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters according to claim 1, characterized in that: Also includes: Calculating the correlation between the spectral index and the chlorophyll content of the rice leaves, and generating a visualization distribution map of the chlorophyll content of the rice leaves in the study area based on the correlation between the spectral index and the chlorophyll content of the rice leaves; and / or, calculating the correlation between the spectral index and the moisture content of the rice plant, and generating a visualization distribution map of the moisture content of the rice plant in the study area based on the correlation between the spectral index and the moisture content of the rice plant; And / or, calculating the correlation between the spectral index and the chlorophyll content of the rice leaves, calculating the correlation between the spectral index and the moisture content of the rice plants, and generating a visualization distribution map of multiple phenotypic parameters of rice in the study area based on the correlation between the spectral index and the severity of rice bacterial leaf blight, the correlation between the spectral index and the chlorophyll content of the rice leaves, and the correlation between the spectral index and the moisture content of the rice plants; the multiple phenotypic parameters include the severity of rice bacterial leaf blight, the chlorophyll content of rice leaves, and the moisture content of rice plants.

3. The method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters according to claim 1, characterized in that: Also includes: Build a sample database; The sample database includes a first sample data set, a second sample data set and a third sample data set; the third sample data set includes multiple sample rice bacterial blight incidence degrees and the sample rice leaf chlorophyll content and sample rice plant moisture content corresponding to each sample rice bacterial blight incidence degree.

4. The method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters according to claim 3, characterized in that: The process of determining the chlorophyll content of the sample rice leaves is as follows: A SPAD-502 chlorophyll meter was used to determine the chlorophyll content of rice leaves corresponding to each plot in the sample area.

5. The method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters according to claim 3, characterized in that: The process of determining the moisture content of the sample rice plants is as follows: The moisture content of the sample rice plants corresponding to each plot in the sample area was calculated using the wet basis moisture content method.

6. A method for predicting the severity of rice bacterial blight based on multiple phenotypic parameters according to claim 1 or 3, characterized in that: The determination process of the first regression model is: The first regression model is determined according to a partial least squares regression method and the first sample data set.

7. A method for predicting the severity of rice bacterial blight based on multiple phenotypic parameters according to claim 1 or 3, characterized in that: The determination process of the second regression model is: The second regression model is determined according to a partial least squares regression method and the second sample data set.

8. The method for predicting the degree of rice bacterial blight disease based on multiple phenotypic parameters according to claim 1, characterized in that: The calculating, based on the multispectral image, the spectral reflectance of rice corresponding to each of the cells specifically includes: Preprocessing the multispectral image; The spectral reflectance of rice is extracted from the preprocessed multispectral image using ENVI software, and then the spectral reflectance of rice corresponding to each plot is determined.

9. A rice bacterial blight disease severity prediction system based on multiple phenotypic parameters, characterized in that: include: A multispectral image acquisition module, used to acquire a multispectral image of rice in a research area; the research area includes a plurality of plots; A rice spectral reflectance calculation module, used for calculating the rice spectral reflectance corresponding to each of the cells based on the multispectral image; A rice leaf chlorophyll content and rice plant moisture content calculation module, used to determine the rice leaf chlorophyll content corresponding to each of the plots based on the rice spectral reflectance and the first regression model, and to determine the rice plant moisture content corresponding to each of the plots based on the rice spectral reflectance and the second regression model; The first regression model is determined based on the first sample data set; the second regression model is determined based on the second sample data set; the first sample data set includes a plurality of sample rice spectral reflectances and a sample rice leaf chlorophyll content corresponding to each of the sample rice spectral reflectances, and the second sample data set includes a plurality of sample rice spectral reflectances and a sample rice plant moisture content corresponding to each of the sample rice spectral reflectances; the sample rice is rice under bacterial blight stress; A rice bacterial blight disease incidence degree determination module is used to determine the rice bacterial blight disease incidence degree corresponding to each of the plots based on the correlation relationship and the rice leaf chlorophyll content and rice plant water content corresponding to each of the plots; the correlation relationship is the correlation relationship between the rice leaf chlorophyll content, the rice plant water content, and the rice bacterial blight disease incidence degree; A spectral index determination module, used for screening the characteristic variables of the rice spectral reflectance based on the score value of the variable projection importance index of the first regression model and the score value of the variable projection importance index of the second regression model to obtain a spectral index; The module for generating a visualization distribution map of the degree of rice bacterial blight disease is used to calculate the correlation between the spectral index and the degree of rice bacterial blight disease, and based on the correlation between the spectral index and the degree of rice bacterial blight disease, generate a visualization distribution map of the degree of rice bacterial blight disease in the study area.

10. The rice bacterial blight disease severity prediction system based on multiple phenotypic parameters according to claim 9, characterized in that: Also includes: A rice leaf chlorophyll content visualization distribution map generation module, used for calculating the correlation between the spectral index and the chlorophyll content of the rice leaves, and generating a rice leaf chlorophyll content visualization distribution map in the study area based on the correlation between the spectral index and the chlorophyll content of the rice leaves; and / or, a rice plant moisture content visualization distribution map generating module, used for calculating the correlation between the spectral index and the rice plant moisture content, and generating a rice plant moisture content visualization distribution map in the study area based on the correlation between the spectral index and the rice plant moisture content; And / or, a rice multi-phenotypic parameter visualization distribution map generation module is used to calculate the correlation between the spectral index and the chlorophyll content of the rice leaves, calculate the correlation between the spectral index and the moisture content of the rice plant, and generate a visualization distribution map of the multi-phenotypic parameters of rice in the study area based on the correlation between the spectral index and the severity of rice bacterial blight, the correlation between the spectral index and the chlorophyll content of the rice leaves, and the correlation between the spectral index and the moisture content of the rice plant; the multi-phenotypic parameters include the severity of rice bacterial blight, the chlorophyll content of rice leaves and the moisture content of rice plants.

Images