Method for constructing a spectral index for monitoring red mottle of chickpea

Abstract

This invention provides a method for constructing spectral indices for monitoring fava bean red spot disease. First, spectral curves of different degrees of fava bean red spot susceptibility are obtained based on spectral reflectance, and the sensitive bands and red spot sensitivity indices for each spectral band are obtained. Then, based on... PROSAIL The model, combined with the erythrophyll sensitivity index of each spectral band, is used to obtain the canopy insensitive bands; finally, the erythrophyll index is constructed. LSDI This method constructs a spectral index that is sensitive to broad bean red spot disease but insensitive to canopy structure, based on the spectral characteristics of broad bean red spot disease and the canopy reflectance characteristics of healthy vegetation. This aims to enhance disease information while suppressing canopy structure difference information, thereby satisfying the requirement that a single spectral feature is applicable to the monitoring of red spot disease in different varieties of broad beans, and providing guidance for broad bean planting and production.

Description

Technical Field

[0001] This invention relates to the field of pest and disease monitoring technology, specifically to a method for constructing a spectral index for monitoring broad bean red spot disease. Background Technology

[0002] Broad beans are the second largest winter crop in the Yangtze River basin of my country, second only to rapeseed in planting area. They are an important source of vegetables for spring supply and a significant source of income for farmers. Broad beans are susceptible to numerous pests and diseases, such as damping-off, red spot, rust, wilt, and American serpentine leafminer. Red spot disease, in particular, can cause large-scale yield reductions and severe economic losses. Currently, identification of red spot disease relies on manual methods, which are inaccurate, time-consuming, and inevitably have a time lag. Furthermore, it is impossible to obtain regional disease incidence data on a macro scale to guide broad bean production.

[0003] Remote sensing technology, which emerged in the 1960s, is a detection technology that, based on electromagnetic wave theory, uses various sensors to collect, process, and image electromagnetic wave information radiated and reflected by distant targets. This comprehensive technology enables the detection and identification of various objects on the ground. Remote sensing can quickly acquire surface spectral information, providing an important technical means for monitoring crop diseases. However, because it is difficult to simultaneously achieve both temporal and spatial resolution in satellite remote sensing, crop disease monitoring based on satellite imagery mainly focuses on extracting the extent of disease occurrence and cannot accurately quantify the severity of diseases.

[0004] In recent years, UAV remote sensing has gained increasing popularity among agricultural researchers due to its advantages such as low cost, high timeliness, low loss, and reusability, which can meet the needs of crops with rapid growth and changes. Based on UAV remote sensing technology, researchers have carried out research in various aspects such as rapid and non-destructive identification, grading assessment, and spatial mapping of crop diseases. For example, in "Remote Sensing Monitoring of Areca Yellowing Disease Based on UAV Multispectral Imagery" (Zhao Jinling et al., Transactions of the Chinese Society of Agricultural Engineering, 2020, 36(08): 54-61), multispectral UAV images of areca were collected based on the RedEdge-M sensor, and the minimum redundancy maximum correlation algorithm was used to select the sensitive features of areca yellowing disease to realize the monitoring of disease severity; in "Wheat yellow rust monitoring by learning from multispectral UAV aerial imagery" (Jinya Su et al., Computers and Electronics in Agriculture, 2018, 155: 157-166), multispectral images of winter wheat were collected by UAV equipped with RedEdge-M sensor, and a wheat stripe rust monitoring model was constructed by random forest method; in "UAV Hyperspectral Monitoring of Maize Canopy Leaf Spot Disease" (Liang Hui et al., Spectroscopy and Spectral Analysis, 2020, 40(06): 1965-1972), the severity of maize leaf spot disease at different growth stages was inverted by constructing UAV hyperspectral index and different levels of disease mapping were realized.

[0005] However, the aforementioned research results indicate that UAV remote sensing can only effectively and quantitatively invert the severity of diseases in a single crop variety. For different varieties of the same crop, different inversion models need to be constructed, and few scholars have studied a single model applicable to remote sensing monitoring of diseases in different crop varieties. Furthermore, in complex planting environments with fragmented plots in mountainous terrain, adjacent plots of the same crop often contain multiple varieties, and the canopy structure of different varieties exhibits subtle differences. The spectral reflectance at the canopy scale includes not only leaf physicochemical parameters but also canopy structure information such as leaf area index (LAI) and leaf inclination angle distribution (LAD). Therefore, existing crop remote sensing monitoring models have low precision and accuracy in hilly and mountainous areas, and cannot effectively provide guidance for broad bean cultivation. Summary of the Invention

[0006] To address the problems existing in the prior art, the present invention aims to provide a method for constructing a spectral index for monitoring broad bean red spot disease. This method constructs a spectral index that is sensitive to broad bean red spot disease but insensitive to canopy structure based on the spectral characteristics of broad bean red spot disease and the canopy reflectance characteristics of healthy vegetation. This achieves the goal of enhancing disease information while suppressing canopy structure difference information, thereby satisfying the requirement that a single spectral feature is applicable to the monitoring of red spot disease in different varieties of broad beans, and providing guidance for broad bean planting and production.

[0007] The objective of this invention is achieved through the following technical solution:

[0008] A method for constructing spectral indices for monitoring fava bean red spot disease, characterized by the following steps: First, obtaining spectral curves of different degrees of fava bean red spot disease susceptibility based on spectral reflectance, and obtaining the sensitive wavelength band [λ] of fava bean red spot disease. a ,λ b [and the susceptibility index of erythroderma for each spectral band; then, based on the PROSAIL model and combined with the susceptibility index of erythroderma for each spectral band, the canopy insensitive band [λ] is obtained. c ,λ d Finally, the LSDI (Low Spot Index) was constructed:

[0009]

[0010] In the formula: k1 and k2 are the coronal insensitive spectral variables, and k1 and k2 change inversely with the severity of the disease;

[0011] in:

[0012]

[0013] In the formula: Indicates wavelength λ c spectral reflectance; Indicates wavelength at λ c+1 -λ d Spectral reflectance within the range;

[0014] Right now:

[0015]

[0016] To further optimize, the specific method for obtaining spectral curves of different degrees of susceptibility to broad bean red spot disease based on spectral reflectance is as follows: ASD HH2 ground cover spectrometer is used to obtain spectral curves of a group of leaves, and the selected group of leaves are leaves with different degrees of susceptibility at the same time.

[0017] The specific method for testing the spectral data of the leaf is as follows: the leaf is clamped at three points on the broad bean leaf, and three spectra are measured at each point. The average value of the nine spectra is taken as the final spectrum of the leaf.

[0018] For further optimization, the sensitive band [λ] a ,λ b [720nm, 950nm].

[0019] For further optimization, the erythroderma sensitivity index is obtained by the standard deviation of the spectral reflectance of each corresponding spectral band; that is, the standard deviation of the spectral reflectance of each spectral band is the erythroderma sensitivity index of the corresponding spectral band.

[0020] To further optimize, the process of obtaining canopy-insensitive bands based on the PROSAIL model and combined with the red spot disease sensitivity index of each spectral band is as follows: LAI (leaf area index) and LAD (leaf tilt angle distribution) of different leaves in the same group are input into the PROSAIL model, while other parameters of the PROSAIL model are fixed to obtain the canopy reflectance spectrum under different LAI and LAD. Combined with the red spot disease sensitivity index of each spectral band for LAI and LAD, the canopy-insensitive bands are obtained.

[0021] For further optimization, the canopy-insensitive band [λ] c ,λ d [780nm, 920nm].

[0022] A monitoring model for spectral indices obtained using the above construction method is characterized by:

[0023] First, use a drone remote sensing platform to collect hyperspectral images of the field;

[0024] Then, on the same day that the field hyperspectral images were collected, the images were divided into multiple square areas of h·h (in meters) as survey plots. The center coordinates of the survey plots were recorded using a handheld GPS. The disease index DI of the survey plot was obtained by examining m broad bean canopy leaves within the survey plot.

[0025]

[0026] In the formula: i represents the disease severity level of a single leaf, which is determined by the percentage of the leaf area covered by the lesion; f(i) represents the total number of leaves in the leaf sample with a disease severity level of i; n represents the highest disease severity level.

[0027] Then, based on the center coordinates of the survey plots recorded by handheld GPS, the corresponding point was found in the remote sensing influence using ENVI software. A square region of interest of h·h (in meters) was constructed with this point as the center, and the average spectral reflectance of all pixels in the region of interest was used as the final spectrum of the survey plot. Then, the LSDI of the corresponding survey plot was obtained using the red spot disease index LSDI constructed by the above method.

[0028] Finally, using the LSDI of the surveyed plots as the independent variable and the DI of the corresponding surveyed plots as the dependent variable, a monitoring model for fava bean red spot disease was established using a univariate quadratic linear regression method:

[0029] y = Ax 2 +Bx-C;

[0030] In the formula: y represents the dependent variable DI; x represents the independent variable LSDI; A, B, and C are constants.

[0031] For further optimization, the drone remote sensing platform adopts the DJI M600 Pro hexacopter drone and is equipped with a PikaL hyperspectral imager.

[0032] For further optimization, the disease severity level value i is divided into: Level 0: No lesions; Level 1: 0-1% with lesions (excluding 0%); Level 2: 1-10% with lesions (excluding 1%); Level 3: 10-20% with lesions (excluding 10%); Level 4: 20-30% with lesions (excluding 20%); Level 5: 30-45% with lesions (excluding 30%); Level 6: 45-60% with lesions (excluding 45%); Level 7: 60-80% with lesions (excluding 60%); Level 8: 80-100% with lesions (excluding 80%).

[0033] The present invention has the following technical effects:

[0034] This application constructs the broad bean red spot disease index LSDI by obtaining measured spectra of diseased leaves and simulated spectra of different canopies, utilizing the spectral characteristics of red spot disease sensitivity but canopy structure insensitivity. For broad bean red spot disease, the LSDI index constructed in this application is compared with existing common spectral indices, and the coefficient of determination R0 is calculated. 2 A score of 0.8 is achieved, which effectively reflects the disease index of broad bean red spot disease. Furthermore, compared to existing spectral indices such as the Enhanced Vegetation Index (EVI) and Anthocyanin Reflectance Index (ARI), which exhibit high correlation, the monitoring model based on the red spot disease index LSDI constructed in this application is less affected by canopy geometry. This significantly improves the monitoring accuracy of toothed leaf spot disease in various broad bean varieties and provides a new methodological reference and guidance for disease monitoring and broad bean red spot disease resistance breeding in areas with fragmented plots, complex planting structures, and diverse varieties. Attached Figure Description

[0035] Figure 1 This is a schematic diagram illustrating different degrees of disease in the same group of leaves in an embodiment of the present invention.

[0036] Figure 2 This is a spectral curve of different disease susceptibility levels of the same group of leaves in an embodiment of the present invention.

[0037] Figure 3 This is a graph showing the sensitivity index curves for different spectrum bands of erythroderma in an embodiment of the present invention.

[0038] Figure 4 This is a graph showing the simulated spectral reflectance of the canopy for different LAIs in this embodiment of the invention.

[0039] Figure 5 This is a graph showing the sensitivity of the canopy spectrum to LAI in an embodiment of the present invention.

[0040] Figure 6 This is a graph showing the relationship between the spectral reflectance ratio N1 and LAI in an embodiment of the present invention.

[0041] Figure 7 This is a graph showing the simulated spectral reflectance of the canopy for different LADs in this embodiment of the invention.

[0042] Figure 8 This is a graph showing the sensitivity of the canopy spectrum to LAD in an embodiment of the present invention.

[0043] Figure 9 This is a graph showing the relationship between the spectral reflectance ratio N1 and LAD in an embodiment of the present invention.

[0044] Figure 10 This is a graph showing the relationship between the LSDI (Likely a type of opioiditis index) and the DI (disease index) in an embodiment of the present invention.

[0045] Figure 11 This is a heatmap showing the correlation coefficients between the spectral index, the red spot disease index, and the disease severity index in an embodiment of the present invention.

[0046] Figure 12 This is a graph showing the functions of the anthocyanin reflectance index (ARI), enhanced vegetation index (EVI), and disease index (DI) in an embodiment of the present invention; wherein: Figure 12 (a) is a graph showing the relationship between the Enhanced Vegetation Index (EVI) and the Disease Index (DI). Figure 12 (b) is a graph showing the relationship between the anthocyanin reflectance index (ARI) and the disease index (DI).

[0047] Figure 13 This is a graph showing the relationship between the predicted and measured values ​​of the monitoring models for the anthocyanin reflectance index (ARI), enhanced vegetation index (EVI), and red spot disease index (LSDI) in this embodiment of the invention; wherein: Figure 13(a) is a graph showing the relationship between the predicted and measured values ​​of the LSDI (Large Spot Index) monitoring model. Figure 13 (b) is a graph showing the relationship between predicted and measured values ​​of the monitoring model for the Enhanced Vegetation Index (EVI). Figure 13 (c) is a graph showing the relationship between the predicted and measured values ​​of the anthocyanin reflectance index (ARI) monitoring model.

[0048] Figure 14 This is an estimated graph of the disease index DI in the test area obtained by the monitoring model of the LSDI index in this embodiment of the invention. Detailed Implementation

[0049] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0050] Example 1:

[0051] A method for constructing a spectral index for monitoring fava bean red spot disease, characterized by the following steps:

[0052] Sample selection before the experiment: Three common varieties were selected: Chenghu No. 10 (susceptible), Tongcanxian No. 8 (resistant), and Doumei No. 1 (moderately susceptible); 20 quadrats were selected for each variety, for a total of 60 quadrats, covering plants with different disease severity (each quadrat was a square area of ​​0.5×0.5m).

[0053] First, spectral curves of different degrees of susceptibility to broad bean red spot disease were obtained based on spectral reflectance: Spectral curves of a group of leaves were acquired using an ASD HH2 ground-based spectrometer (spectral curves of the same group of leaves are shown in the figure). Figure 2 As shown), the selected set of leaves consisted of leaves from the same period but with different degrees of disease susceptibility (e.g., leaves from the same period but with different degrees of disease susceptibility). Figure 1 As shown: Five leaves (C1-C5) of different susceptibility levels at the same time were selected as the same group of leaves. Figure 1 The severity of the disease gradually increases from C1 to C5.

[0054] The specific method for testing the spectral data of the leaf is as follows: the leaf is clamped at three points on the broad bean leaf, and three spectra are measured at each point. The average value of the nine spectra is taken as the final spectrum of the leaf.

[0055] like Figure 2As shown, due to the presence of strong pigment absorption bands in plant leaves, leaves of different disease severities exhibit low spectral reflectance in the visible light band of 390nm–700nm. Near 550nm, a distinct reflection peak is formed due to chlorophyll reflecting green light; and near 680nm, an absorption trough is formed due to chlorophyll absorbing red light. Clearly, the more severe the disease, the greater the degree of chlorophyll dissolution and decomposition within the leaf cells, resulting in weaker reflection of green light and weaker absorption of red light, but overall the difference is not significant. In the near-infrared spectral band of 720–950nm, influenced by changes in leaf cell structure, the spectral reflectance of leaves of different disease severities shows significant differences, exhibiting a certain pattern: leaves of more severe disease have lower spectral reflectance, and the difference in spectral reflectance decreases with increasing wavelength.

[0056] Based on the above analysis, the sensitive wavelength range of 720–950 nm for broad bean red spot disease and the red spot disease sensitivity index for each spectral band were obtained. The red spot disease sensitivity index was obtained by the standard deviation of the spectral reflectance of each corresponding spectral band (i.e., the standard deviation of the spectral reflectance of each spectral band is the red spot disease sensitivity index for the corresponding spectral band); the red spot disease sensitivity index curves for each spectral band were obtained, as shown in the figure. Figure 3 As shown. Figure 3 In the visible spectrum, the sensitivity index for fava bean red spot disease is below 0.03, indicating low sensitivity and insufficient for monitoring the disease. The near-infrared band (720–950 nm) shows significantly higher sensitivity for red spot disease than the visible spectrum, especially within the 750–780 nm range, where the sensitivity index is greater than 0.1, making it a suitable band for monitoring fava bean red spot disease. The sensitivity peaks at 760 nm, and then gradually decreases with increasing wavelength. This further confirms that the sensitive band for fava bean red spot disease is 720–950 nm.

[0057] Then, based on the PROSAIL model and combined with the erythroderma sensitivity index of each spectral band, the canopy insensitive band [λ] was obtained. c ,λ d Specifically:

[0058] Based on historical experimental data, the measured LAI values ​​in the experimental field range from 4.1 to 6.6. In this embodiment, based on the PROSAIL model, the model parameter LAI was set to 2–8 with a step size of 0.1, while keeping other model parameters consistent. Sixty-one canopy spectra were simulated, and spectral curves with integer LAI values ​​were selected to obtain spectral curve plots (seven spectral curves in this embodiment, such as…). Figure 4 (As shown). Figure 4In the 400–500 nm and 580–710 nm ranges, the spectral reflectance decreases with increasing leaf area, but the magnitude is not significant. In the near-infrared band of 720–1050 nm, the canopy spectral reflectance increases significantly with increasing leaf area. Moreover, in the 780–920 nm band, the canopy spectral reflectance for the same leaf area remains almost unchanged, and the spectral curve is approximately a straight line.

[0059] For the 61 simulated canopy spectra with different leaf areas, the susceptibility index of red spot disease in each spectral band was calculated, and the results are as follows: Figure 5 As shown. Figure 5 In the visible spectrum range of 400–710 nm, the sensitivity index of canopy spectral reflectance to leaf area for red spot disease is less than 0.005, indicating that the influence of different leaf areas on canopy spectral reflectance is relatively weak in this range. However, in the near-infrared band, the sensitivity index gradually increases to above 0.035, and the influence of leaf area on canopy spectral reflectance becomes increasingly significant. The ratio of spectral reflectance at 920 nm to 780 nm for the canopy spectra of the above 61 different leaf areas was calculated respectively. (where R is in the formula) 920 Represents the spectral reflectance at 920 nm, R 780 (Representing the spectral reflectance at 780 nm), the relationship between the ratio N1 and LAI is obtained, such as... Figure 6 As shown; Figure 6 The scatter plots represent the canopy spectral index N1 for different leaf areas, and the straight line is the fitted line from the scatter plots. Figure 6 It can be seen that N1 < 1.05; when LAI > 4, N1 is close to 1; and the maximum value of the scatter plot is 1.034, the minimum value is 1.002, the mean is 1.01, the standard deviation is 0.0084, and the angle between the fitted line and the line y = 1 is 0°14'47". Therefore, N1 is not sensitive to leaf area, and this index can reduce the influence of leaf area on canopy spectral reflectance.

[0060] Meanwhile, based on historical experimental data, it is known that the leaf tilt angle of the broad bean canopy leaves in the experimental field ranges from 30° to 70°. In this embodiment, based on the PROSAIL model, the leaf tilt angle range is set to 30° to 80° with a step size of 1°. Maintaining consistency in other model parameters, 51 canopy reflectance spectra under different leaf tilt angles are simulated. Spectral curves with leaf tilt angles that are integer multiples of 10° are selected to obtain spectral curve diagrams (6 spectral curves in this embodiment, such as...). Figure 7 (As shown). Figure 7In the visible light bands of 400–500 nm and 650–700 nm, the canopy reflectance does not differ much under different leaf tilt angles; in the near-infrared band of 720–1050 nm, the canopy spectral reflectance decreases significantly with increasing leaf tilt angle, and in the spectral band of 780–920 nm, the canopy spectral reflectance at the same leaf tilt angle remains almost unchanged, and the spectral curve is approximately a straight line.

[0061] For the 51 simulated canopy spectra with different leaf areas, the susceptibility index of red spot disease in each spectral band was calculated, and the results are as follows: Figure 8 As shown. Figure 8 In the 400–700 nm spectral band, the sensitivity index of canopy spectral reflectance to leaf tilt angle for red spot disease is less than 0.03, especially in the 400–500 nm and 650–700 nm bands where the sensitivity index is less than 0.01, indicating that the canopy spectrum is relatively less affected by the leaf area angle (LAD). In the near-infrared band of 750–1050 nm, the sensitivity index for red spot disease is greater than 0.1, indicating that canopy spectral reflectance is closely related to the leaf area angle (LAD) in this band. The ratio of spectral reflectance at 920 nm to 780 nm for the 51 different leaf area canopy spectra was calculated respectively. (where R is in the formula) 920 Represents the spectral reflectance at 920 nm, R 780 (Representing the spectral reflectance at 780 nm), the relationship between the ratio N1 and LAD is obtained, such as... Figure 9 As shown; Figure 9 The scatter plots represent the canopy spectral index N1 for different leaf areas, and the straight line is the fitted line from the scatter plots. Figure 9 It can be seen that N1 < 1.05; when LAD < 60°, N1 is close to 1; and the maximum value of the scatter plot is 1.044, the minimum value is 1.006, the mean is 1.013, the standard deviation is 0.01, and the angle between the fitted line and the line y = 1 is 0°1'57", close to 0°. Therefore, N1 is not sensitive to leaf tilt angle, and this index can reduce the influence of leaf tilt angle on canopy spectral reflectance.

[0062] In summary, the canopy-insensitive band is 780–920 nm; the ratio spectral index in this band can effectively suppress spectral noise caused by differences in leaf area and leaf tilt angle.

[0063] Finally, the LSDI index for erythroderma was constructed:

[0064]

[0065] In the formula: k1 and k2 are the coronal insensitive spectral variables, and k1 and k2 change inversely with the severity of the disease;

[0066] in:

[0067]

[0068] In the formula: Indicates wavelength λ c spectral reflectance; Indicates wavelength at λ c+1 -λ d Spectral reflectance within the range;

[0069] Right now:

[0070]

[0071] Based on the above analysis, the canopy insensitive wavelength range in this embodiment is 780–920 nm; therefore, the LSDI (Laminated Spot Index) is:

[0072]

[0073] For R 781-920 In this embodiment, the value of R is selected from the 910nm spectral band, which has relatively low sensitivity to red spot disease (of course, R...). 781-920 Alternatively, it can be the average spectral reflectance of each band in that band; therefore, the final LSDI (Likely referring to a specific index of erythroderma) is:

[0074]

[0075] Example 2:

[0076] A monitoring model for spectral indices obtained using the construction method in Example 1, characterized in that:

[0077] First, a drone remote sensing platform was used to collect hyperspectral images of the field; the drone remote sensing platform used a DJI M600 Pro hexacopter drone and was equipped with a Pika L hyperspectral imager.

[0078] The weather was clear and cloudless, and hyperspectral image acquisition in the field was conducted between 11:00 and 13:00, when the solar altitude angle was at its maximum. The UAV flew at an altitude of 60 meters, a speed of 2 m / s, a lateral overlap of 60%, and a camera frame rate of 80. Simultaneously, the acquired image data underwent preprocessing including geometric correction, radiometric correction, and spectral smoothing (using existing conventional techniques).

[0079] Then, on the same day that the field hyperspectral images were collected, the images were divided into multiple square areas of h·h (unit: meters, h = 0.5 in this example) as survey plots. Three common varieties were selected: Chenghu 10 (susceptible), Tongcanxian 8 (resistant), and Doumei 1 (moderately susceptible). Twenty survey plots were selected for each variety, for a total of 60 survey plots, covering plants with different disease severity. The center coordinates of the survey plots were recorded using a handheld GPS, and the disease index DI of the survey plot was obtained by examining m leaves of broad bean canopy within the survey plot (m = 50 in this example).

[0080]

[0081] In the formula: i represents the disease severity level of a single leaf, determined by the percentage of lesion area to leaf area; the disease severity level i is specifically divided into: Level 0: no lesions; Level 1: 0-1% with lesions (excluding 0%); Level 2: 1-10% with lesions (excluding 1%); Level 3: 10-20% with lesions (excluding 10%); Level 4: 20-30% with lesions (excluding 20%); Level 5: 30-45% with lesions (excluding 30%); Level 6: 45-60% with lesions (excluding 45%); Level 7: 60-80% with lesions (excluding 60%); Level 8: 80-100% with lesions (excluding 80%). f(i) represents the total number of leaves in the leaf sample with a disease severity level of i; n represents the highest disease severity level, in this embodiment n = 8.

[0082] Then, based on the center coordinates of the survey plot recorded by the handheld GPS, the corresponding point was found in the remote sensing influence using ENVI software. A square region of interest of h·h (in meters, h = 0.5 in this embodiment) was constructed with this point as the center. The average spectral reflectance of all pixels in the region of interest was used as the final spectrum of the survey plot. Then, the LSDI of the corresponding survey plot was obtained using the red spot disease index LSDI constructed by the method in Example 1.

[0083] Finally, 15 quadrats for each variety, totaling 45 quadrats, were selected as the modeling sample, and the remaining 15 quadrats were used as the validation sample. Using the LSDI of the quadrats as the independent variable and the DI of the corresponding quadrats as the dependent variable, a monitoring model for broad bean red spot disease was established using a univariate quadratic linear regression method.

[0084] y = Ax 2 +Bx-C;

[0085] In the formula: y represents the dependent variable DI; x represents the independent variable LSDI; A, B, and C are constants;

[0086] In this embodiment: y = -7.35x 2+7.057x-0.08049.

[0087] Example 3:

[0088] A method for validating the LSDI index for erythema in Example 1 and the monitoring model for Example 2, characterized in that:

[0089] First, Pearson correlation analysis was performed on the LSDI index of red spot disease obtained by 13 common spectral indices in this field and the disease index DI of the sample obtained by the method in Example 1, and the correlation coefficient heatmap was obtained. Figure 11 As shown in Table 1, 13 common spectral indices are listed below:

[0090] Table 1:

[0091]

[0092] Depend on Figure 11 It can be seen that all 14 spectral indices are associated with favism, with LSDI, EVI, and ARI showing significant correlations with favism, with correlation coefficients of 0.84, -0.80, and 0.79, respectively. Therefore, using the above three spectral indices as independent variables and the disease index as the dependent variable, a univariate quadratic linear regression method was used to establish a favism disease monitoring model for a modeling sample set containing 45 quadrats. The results are as follows. Figure 10 and Figure 12 As shown. Model R based on the LSDI index. 2 The LSDI index reached 0.8, the highest of the three, while RSME was the lowest, indicating that the LSDI index was relatively more effective in retrieving the severity index of fava bean red spot disease.

[0093] A disease monitoring model for broad bean leaf spot disease was used, incorporating the anthocyanin reflectance index (ARI), enhanced vegetation index (EVI), and leaf spot disease index (LSDI), to predict the disease index of 15 independent survey samples. The predicted values ​​were then compared with the measured values. The comparison results are as follows: Figure 13 As shown in the figure, the dashed lines are 1:1 contour lines, and the solid lines represent the fitting lines between the measured and predicted DI values. Figure 13 It can be seen that the slope of the regression line between the predicted and measured DI values ​​of the LSDI model is closer to 1, and the intercept is closer to 0; in terms of accuracy, the R-squared value of the LSDI model is... 2 The LSDI index was 0.84, which was 0.05 and 0.12 higher than the EVI and ARI models, respectively; the RMSE of the LSDI model was 0.043, which was 0.005 and 0.013 lower, respectively. Therefore, the LSDI index can improve the prediction accuracy of the disease index of fava bean red spot disease, and the regression model with LSDI as the independent variable is the best estimation model.

[0094] The LSDI model was applied to the experimental area to obtain an estimated DI value map of broad bean canopy red spot disease in the experimental area, as shown in the figure. Figure 14 As shown.

[0095] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims. Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This method of description is merely for clarity, and those skilled in the art should consider the specification as a whole. The technical solutions in the various embodiments can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for constructing a spectral index for monitoring red mottle of Vicia faba, characterized by: Specifically, the following steps are included: Firstly, the spectral curves of different degrees of chickpea red spot disease were obtained based on spectral reflectance, and the sensitive band of red spot disease was obtained and the red spot disease sensitive index of each spectral segment. Then, based on PROSAIL The model, combined with the erythrofoam sensitivity index of each spectral band, obtains the canopy insensitive bands. Specifically, this involves: separating different leaves from the same group. LAI and LAD enter PROSAIL Model, while fixed PROSAIL Other parameters of the model yield different results. LAI and LAD The lower canopy reflectance spectrum, combined with the effect of each spectral band on LAI and LAD The sensitivity index for okra was used to obtain the canopy-insensitive band; Finally, the rust index was constructed LSDI : ; wherein: and respectively are the canopy insensitive spectral variables and change inversely with the degree of disease severity, and respectively. in: ， ； In the formulae: denotes the spectral reflectance of light having a wavelength of denotes the spectral reflectance of light having a wavelength in the range of denotes the spectral reflectance of light having a wavelength in the range of​ Right now: 。 2. The method according to claim 1, wherein the method is characterized by: The specific method for obtaining spectral curves of different degrees of susceptibility to broad bean red spot disease based on spectral reflectance is as follows: ASD HH2 ground cover spectrometer is used to obtain spectral curves of a group of leaves, and the selected group of leaves are leaves with different degrees of susceptibility at the same time. The specific method for testing the spectral data of the leaf is as follows: the leaf is clamped at three points on the broad bean leaf, and three spectra are measured at each point. The average value of the nine spectra is taken as the final spectrum of the leaf.

3. A method for constructing a spectral index for monitoring fava bean red spot disease according to claim 1 or 2, characterized in that: The erythroderma sensitivity index is obtained by the standard deviation of spectral reflectance for each corresponding spectral band.

4. A method for constructing a monitoring model using the spectral indices obtained by the method for constructing spectral indices for monitoring fava bean red spot disease as described in claim 1, characterized in that: First, use a drone remote sensing platform to collect hyperspectral images of the field; Then, on the same day that the field hyperspectral images were collected, the field hyperspectral images were divided into multiple... h·h A square area was selected as the survey sample plot, and a handheld device was used. GPS Record the coordinates of the center of the survey plot and the leaves of the broad bean canopy within the survey plot. m Zhang obtained the disease index of the survey sample. DI : ； In the formula: i The disease severity score for a single leaf is determined by the percentage of the leaf area covered by the lesion. f(i) The disease severity value in the leaf sample is indicated by i The total number of leaves; n This indicates the highest severity level of the illness; Afterwards, based on the handheld GPS The recorded coordinates of the center of the survey quadrat were used. ENVI The software finds the corresponding point in the remote sensing imagery and constructs a system centered on that point. h·h A square region of interest is defined, and the average spectral reflectance of all pixels within that region is used as the final spectrum of the survey plot. Then, the red spot disease index is constructed using the method described above. LSDI Obtain the corresponding survey sample LSDI ; Finally, the investigation plots were surveyed LSDI As the independent variable, the corresponding investigation plots were surveyed DI As the dependent variable, the monitoring model of the bean red spot disease was established by the method of one-dimensional quadratic linear regression: ； wherein: y represents the dependent variable DI ; x represents the independent variable LSDI ; A , B , C are constants, respectively.

5. The method of claim 4, wherein the method of constructing a monitoring model of the spectral index obtained by the method of constructing a spectral index for monitoring red mottle of Vicia faba is characterized by: The drone remote sensing platform uses the DJI M600 Pro hexacopter drone and is equipped with a Pika L hyperspectral imager.

6. The method for constructing a monitoring model based on the spectral indices obtained by the method for constructing spectral indices for monitoring fava bean red spot disease according to claim 4, characterized in that: The disease severity value i The grades are divided into: Grade 0: No lesions; Grade 1: 0-1% with lesions; Grade 2: 1-10% with lesions; Grade 3: 10-20% with lesions; Grade 4: 20-30% with lesions; Grade 5: 30-45% with lesions; Grade 6: 45-60% with lesions; Grade 7: 60-80% with lesions; Grade 8: 80-100% with lesions.

Images