A method for constructing and estimating an index of photosynthetic capacity of a maize population

Abstract

The application discloses a kind of corn population photosynthetic capacity index construction and estimation method, it is related to the field of agricultural products monitoring, comprising the following steps: S1: obtaining corn Pn, Gs, Ci and Tr four photosynthetic parameters, S2: using Akima spline curve interpolation method Pn, Gs, Ci and Tr measurement interval is shortened to one hour, S3: the fitting curve of four parameters is obtained, S4: the projected area under fitting curve is calculated, S5: the projected area is normalized, determine the weight of each projected area, S6: calculate the new index ODP of photosynthetic capacity, while providing ODP index feasibility verification method and estimation method.The application has the advantages that, more accurately reflect population photosynthetic capacity, consistent with the actual situation, can truly reflect the growth state of field corn.In addition, using unmanned aerial vehicle hyperspectral can quickly and accurately estimate the index, improve the efficiency of photosynthetic capacity identification.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crop monitoring, and particularly relates to a method for constructing and estimating a corn population photosynthetic capacity index. BACKGROUND

[0002] Corn is an important cereal crop, which has food, feed and various industrial uses, and has an important strategic position in ensuring food security. In addition, corn is a typical C4 plant, which has stronger photosynthesis than C3 plants, which also indicates that photosynthesis has a greater impact on corn yield. Photosynthesis provides the material basis for grain formation, and its strength is an important indicator for measuring crop growth conditions. Therefore, monitoring the photosynthesis of corn is of great significance for understanding its growth conditions.

[0003] Traditional evaluation of crop photosynthetic capacity mainly relies on gas exchange method determination, however, this measurement method is time-consuming and laborious, and is not suitable for high-throughput crop photosynthesis research. Multi / hyperspectral sensors can be carried on ground observation platforms, unmanned aerial vehicles or satellite remote sensing platforms to obtain vegetation parameters, so that they are more convenient and faster to collect, and have significant advantages in extracting photosynthetic parameters in the field. The quantitative model between photosynthetic index and spectral vegetation index can quickly and non-destructively obtain the physiological state of vegetation, accurately reflect the physiological changes of vegetation photosynthesis, and effectively estimate the photosynthetic capacity of vegetation.

[0004] Many indicators can represent photosynthetic capacity, such as maximum carboxylation rate, maximum electron transport rate, maximum carbon dioxide (CO2) assimilation rate, however, the determination of these indicators is usually carried out under specific conditions, including optimal light intensity, sufficient CO2 supply and suitable temperature, etc., which mainly reflects the maximum capacity of a specific step in photosynthesis. In the actual field environment, these conditions cannot be met at the same time. Net photosynthetic rate (Pn), stomatal conductance (Gs), intercellular CO2 concentration (Ci) and transpiration rate (Tr) can also represent photosynthetic capacity. Among them, Fn refers to the amount of carbon dioxide absorbed by plant photosynthesis minus the amount of carbon dioxide released by respiration under actual environmental conditions; Gs refers to the degree of opening of the stomata on the plant leaves, which determines the acquisition of O2, CO2 and the loss of water; Ci refers to the concentration of CO2 in the cells, which is one of the important factors affecting the rate of photosynthesis; Tr refers to the amount of water lost by the plant through the stomata, and appropriate transpiration helps to maintain water balance and temperature regulation in the plant body. Each parameter provides a different perspective and information, and these parameters interact with each other to determine the photosynthetic state of the plant. Therefore, when evaluating the photosynthetic capacity of corn using remote sensing means, a single parameter such as Pn cannot accurately represent the photosynthetic capacity of the population, and a comprehensive indicator needs to be constructed to increase the robustness of the evaluation of photosynthetic capacity. In addition, Fn, Gs, Ci and Tr all reflect the instantaneous measurement value at that time, which may be affected by short-term environmental changes, leading to data deviation, which limits the comprehensive evaluation of plant growth and carbon fixation capacity. In view of this, the present application provides a corn population photosynthetic capacity index construction and estimation method. SUMMARY

[0005] The purpose of the present application is to solve the problems in the prior art and to provide a corn population photosynthetic capacity index construction and estimation method.

[0006] A corn population photosynthetic capacity index construction and estimation method, comprising the following steps:

[0007] S1: Obtain the four photosynthetic parameters of corn Pn, Gs, Ci and Tr, and measure every two hours from 6 am to 6 pm.

[0008] S2: Use Akima spline interpolation method to shorten the measurement interval of Pn, Gs, Ci and Tr to one hour.

[0009] S3: Fit the Pn, Gs, Ci and Tr values using the sine and function to obtain the fitting curve of the four parameters, and the sine and function formula is:

[0010]

[0011] Wherein, a is the amplitude, b is the frequency, and c is the phase constant of each sine wave. n is the number of items in the sequence, 1≤n≤8.

[0012] S4: Calculate the projected area under the curve using the integral method, get 、 、 and .

[0013] S5: Normalize the projected area, and determine the weight of each projected area using the CRITIC weight method.

[0014] S6: According to the projected area and weight of each parameter, calculate the photosynthesis new index ODP (One day photosynthesis);

[0015] S7: Get the 100-grain weight of the seed, simulate the corn filling process, and measure the feasibility of the ODP index by the increase of the 100-grain weight.

[0016] S8: Obtain the hyperspectral image of the target area.

[0017] S9: Preprocess the hyperspectral image of the target area, and convert the grid image of the target area to hyperspectral reflectivity.

[0018] S10: Calculate 17 vegetation indices, and use two-dimensional correlation spectral analysis technology to screen sensitive bands and optimize and improve vegetation indices.

[0019] S11: Use competitive adaptive reweighted sampling (CARS) to screen vegetation indices and remove multicollinearity.

[0020] S12: Construct an ODP estimation model based on vegetation indices.

[0021] S13: The coefficient of determination (R 2 ), root mean square error (RMSE) and mean absolute error (MAE) are used to evaluate the estimation performance of the model, and the R 2 , RMSE and MAE calculation formulas are as follows:

[0022]

[0023] Wherein, is the number of samples, and are the measured ODP and estimated ODP, respectively, is the average value of the measured ODP.

[0024] In the above method for constructing and estimating the photosynthetic capacity index of corn population, in step S1, natural light is used as the light source, 3 corn plants with consistent growth conditions, no pests, no damage are randomly selected, and the ear leaf is selected for determination. Keep the leaf in its original position during measurement. Calculate the average value and use it for subsequent data analysis.

[0025] In the above method for constructing and estimating the photosynthetic capacity index of a corn population, in step S5, the projected area is normalized, and the normalized interval is [0, 1], and the CRITIC weight formula is:

[0026]

[0027] wherein, is the intra-index variability, is the inter-index conflictivity, is the index mean, is the correlation coefficient between the indexes i and j, represents the information amount, is the maximum weight.

[0028] In the above method for constructing and estimating the photosynthetic capacity index of a corn population, in step S6, the ODP calculation formula is:

[0029]

[0030] wherein, , , and are the weights of , , and respectively.

[0031] In the above method for constructing and estimating the photosynthetic capacity index of a corn population, in step S7, 30 plants with the same growth vigor are selected and marked at the silking stage of the corn. At 10, 20, 30 and 40 days after flowering, 5 corn ears are selected respectively, and 100 seeds are taken from the middle of each ear. The seeds are dried in an oven at 105°C for 30 minutes, and then further dried at 80°C until the weight is constant. The weight of 500 seeds is measured with an electronic scale, and the hundred-grain weight (HGW) is calculated. The grain filling process is simulated by a Logistic equation, and the HGW measured every 10 days after flowering is taken as the dependent variable. The Logistic equation is:

[0032]

[0033] wherein, t is the number of days after flowering, A is the final hundred-grain weight, B is the initial value parameter, and C is the growth rate parameter.

[0034] In the above method for constructing and estimating the photosynthetic capacity index of the corn population, in step S8, the flight height of the unmanned aerial vehicle is 30 m, the heading overlap rate and the lateral overlap rate are 70% and 60% respectively, before the unmanned aerial vehicle takes off, a standard white board is used for reflectivity calibration. After the unmanned aerial vehicle takes off, gray cloth with reflectivity of 20%, 50% and 70% is used for atmospheric calibration. The image acquisition adopts an automatic exposure mode, and the hyperspectral image acquisition time is at 11:00 on the day of photosynthetic parameter measurement.

[0035] In the above method for constructing and estimating the photosynthetic capacity index of the corn population, in step S9, after image acquisition, the image is calibrated for lens, reflectivity and atmosphere. The calibrated image is spliced by using a position and attitude measurement system (POS), latitude and longitude data and feature point matching of the image overlap area to obtain a hyperspectral raster image containing 176 bands. Finally, the canopy reflectivity of each plot is extracted. The Savitzky-Golay method is used to smooth the reflectivity curve, and the window size is set to 5.

[0036] In the above method for constructing and estimating the photosynthetic capacity index of the corn population, in step S11, the sampling frequency is set to 100000, and the wavelength variable subset corresponding to the minimum root mean square error is selected as the characteristic wavelength.

[0037] In the above method for constructing and estimating the photosynthetic capacity index of the corn population, in step S12, the gradient method is used to divide the training set and the test set in a ratio of 3:1. The specific operation is as follows: first, arrange the samples from small to large according to the ODP value, then select the second or third sample in every four samples as the test set, and the remaining samples as the training set, ensure that the ODP value of the test set samples is within the ODP value range of the training set samples, and the distribution is uniform, and use leave-one-out cross validation (LOOCV) to further evaluate the estimation performance of the model.

[0038] Compared with the prior art, the present application has the following advantages:

[0039] 1. The present application proposes a new photosynthetic index ODP for describing the photosynthesis of corn in a day by measuring the changes of photosynthetic parameters Pn, Gs, Ci and Tr in a day, and further predicts ODP using an unmanned aerial vehicle hyperspectral image, in order to quickly evaluate the photosynthetic capacity of the corn population.

[0040] 2. A comprehensive index is constructed to increase the robustness of photosynthetic capacity evaluation, avoid the problem of data deviation caused by short-term environmental changes, and improve the comprehensive evaluation of plant growth and carbon fixation capacity. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 Flow chart of the method for constructing and estimating the photosynthetic capacity index of the population.

[0042] Figure 2 Pn fitting curve projection area calculation process diagram;

[0043] Figure 3 Maize kernel filling dynamic process diagram;

[0044] Figure 4 Correlation analysis diagram of photosynthetic parameters and hundred-grain weight at different times after flowering;

[0045] Figure 5 Two-dimensional correlation spectrum analysis result diagram;

[0046] Figure 6 ODP and vegetation index correlation analysis diagram;

[0047] Figure 7 Comparison diagram of LOOCV results of different VI screening methods. DETAILED DESCRIPTION

[0048] REFERENCE Figures 1-7 A method for constructing and estimating a corn population photosynthetic capacity index, comprising the following steps:

[0049] S1: Use Li-6400 XT portable photosynthesis system to collect photosynthetic data (including Pn, Gs, Ci and Tr) of corn at 10, 20, 30 and 40 days after flowering. Natural light is used as the light source, and measurements are taken every two hours from 6 am to 6 pm. Randomly select 3 corn plants with consistent growth conditions, no pests and no damage, and select the ear leaf blades for measurement. Keep the leaf blades in their original position during measurement. Calculate the average value and use it for subsequent data analysis.

[0050] S2: Use Akima spline curve interpolation method to shorten the measurement interval of Pn, Gs, Ci and Tr to one hour.

[0051] S3: Use sine and function to fit Pn, Gs, Ci and Tr values to obtain the fitting curves of the four parameters. The sine and function formula is:

[0052]

[0053] Where a is the amplitude, b is the frequency, and c is the phase constant of each sine wave. n is the number of terms in the sequence, 1 ≤ n ≤ 8.

[0054] S4: Use the integral tool of Origin2021 software to calculate the projection area under the fitting curve, and obtain , , and .

[0055] S5: normalizing the projection area, the normalized interval is [0, 1], and the weight of each projection area is determined by using the CRITIC weight method. The formula of the CRITIC weight method is:

[0056]

[0057] wherein, is the intra-variability of the index, is the inter-conflict of the index, is the mean of the index, is the correlation coefficient between the index i and j, represents the information amount, is the maximum weight.

[0058] S6: according to the projection area and weight of each parameter, the photosynthesis new index ODP (One day photosynthesis) is calculated, and the formula of ODP is:

[0059]

[0060] wherein, , , and are the weights of , , and respectively.

[0061] S7: the hundred-grain weight of the seed is obtained, and the corn filling process is simulated, and the feasibility of the ODP index is measured by the increase of the hundred-grain weight. Specifically, 30 plants with the same growth trend are selected at the silking stage of corn and marked. At 10, 20, 30 and 40 days after flowering, 5 corn ears are selected respectively, and 100 seeds are taken from the middle of each ear. The seeds are killed in an oven at 105°C for 30 minutes, and then further dried at 80°C for 24 hours. The weight of 500 seeds is measured by an electronic scale with an accuracy of 0.001g, and the hundred-grain weight (HGW) is calculated. The filling process of the grain is simulated by using the Logistic equation, and the days after flowering are used as the independent variable, and the HGW measured every 10 days is used as the dependent variable. The Logistic equation is:

[0062]

[0063] wherein, t is the days after flowering, A is the final hundred-grain weight, B is the initial value parameter, and C is the growth rate parameter.

[0064] S8: The hyperspectral images of the target area were acquired using DJI M600 Pro with GaiaSky-mini hyperspectral imager. Before image acquisition, route planning and waypoint design were performed using DJI GS Pro. The flight height of the UAV was 30 m, and the lateral and lateral overlap rates were 70% and 60%, respectively. Before the UAV took off, a standard white plate was used for reflectance calibration. After the UAV took off, gray cloths with reflectance of 20%, 50%, and 70% were used for atmospheric calibration. The image acquisition was performed in automatic exposure mode, and the hyperspectral image acquisition time was at 11:00 am on the day of photosynthetic parameter measurement.

[0065] S9: The hyperspectral images of the target area were preprocessed. Image calibration was performed using SpecView software, and the calibration process included lens calibration, reflectance calibration, and atmospheric calibration. The calibrated images were stitched using a position and attitude measurement system, latitude and longitude data, and feature point matching of the image overlap area to obtain hyperspectral raster images containing 176 bands. The raster images of the target area were converted to hyperspectral reflectance using ENVI 5.6 software. The Savitzky-Golay method was used to smooth the reflectance curve, and the window size was set to 5.

[0066] S10: Seventeen vegetation indices (VIs) were calculated, and sensitive bands were selected using two-dimensional correlation spectral analysis techniques to optimize and improve vegetation indices. The calculation formula of the vegetation indices is:

[0067]

[0068] NIR, RedE, R, and G represent the reflectance of near-infrared, red edge, red, and green bands, respectively.

[0069] S11: Competitive adaptive reweighted sampling (CARS) was used to select vegetation indices, and the number of Monte Carlo samples was set to 100000. The wavelength variable subset corresponding to the minimum root mean square error was selected as the characteristic wavelength.

[0070] S12: ODP estimation models were constructed based on vegetation indices. The gradient method was used to divide the training set and test set in a ratio of 3:1. The specific operation was as follows: first, arrange the samples from small to large according to the ODP value, then select the 2nd or 3rd sample in every 4 samples as the test set, and the remaining samples as the training set, ensuring that the ODP value of the test set samples is within the ODP value range of the training set samples and is uniformly distributed. Leave-one-out cross-validation (LOOCV) was used to further evaluate the estimation performance of the model.

[0071] S13: The coefficient of determination (R 2 ), root mean square error (RMSE), and mean absolute error (MAE) were used to evaluate the estimation performance of the model.

[0072] The R 2 , the calculation formula is:

[0073]

[0074] The RMSE calculation formula is:

[0075]

[0076] The MAE calculation formula is:

[0077]

[0078] wherein, is the number of samples, and are the measured ODP and the estimated ODP, respectively, is the average value of the measured ODP.

[0079] A flow chart of a population photosynthetic capacity index construction and estimation method is shown in Figure 1 , including data acquisition, hyperspectral image processing, new index construction, and grain filling process simulation.

[0080] Figure 2 a shows the change of Pn during the day on the 10th day after flowering. With the increase of light and temperature after sunrise, Pn gradually rises and reaches a peak at 10 am. At noon, due to the strong transpiration caused by high temperature, part of the stomata closes, reducing the supply of CO2, and Pn decreases at 12 pm. In the afternoon, with the decrease of light and temperature, the stomata reopen, the supply of CO2 increases, and Pn rises at 14 pm. After that, the light and temperature continue to decrease, and Pn decreases to zero after sunset. The Akima method is used to interpolate the Pn data during the day, and the results are shown in Figure 2 b. Then, the Pn values are fitted using the sine and cosine function. When the number of terms is 5, the decline trend at noon can be accurately fitted ( Figure 2 c), and finally, the projection area of the fitted curve on the x-axis is calculated ( Figure 2 d), and the shaded part in the figure represents the net photosynthesis of corn in a day. The larger the projection area, the more organic matter produced by photosynthesis. Similarly, the changes of Gs, Ci and Tr during the day are fitted. The R2 of all the fittings ranges from 0.95 to 1, achieving high fitting accuracy.

[0081] To comprehensively utilize Pn_area, Gs_area, Ci_area, and Tr_area, the CRITIC weighting method was used to calculate the weight of each indicator, and the results are shown in Table 1. The weight distribution of each indicator shows significant differences. The weights of Pn_area at 10, 20, 30, and 40 days after flowering are 28.241%, 32.747%, 25.478%, and 24.844%, respectively, showing an initial upward trend followed by a downward trend as the grouting process progresses. The weight of Tr_area consistently exceeds 25% and shows a gradual upward trend. Among the four indicators, Pn_area has the highest weight at 10 and 20 days after flowering, while Tr_area has the highest weight at 30 and 40 days after flowering. The weights of Gs_area and Ci_area are relatively low, with Ci_area having a weight of only 14.831% at 20 days after flowering. Finally, the ODP for each grouting stage was calculated.

[0082] Table 1. Weights of different photosynthetic parameters at different grouting stages

[0083]

[0084] Figure 3 A shows the 100-grain weight results at different grain-filling stages. At 10 and 20 days after flowering, the differences between plots were small, with standard deviations of 0.639 and 1.045, respectively. At 30 and 40 days after flowering, and after harvest, the differences between plots were larger, with standard deviations of 2.673, 2.198, and 2.391, respectively. After harvest, the maximum 100-grain weight was 34.508 g, the minimum was 25.355 g, and the average was 30.498 g. Figure 3 b shows the weight of 100 grains in a certain plot in 2021. The dynamics of corn grain filling were fitted using the Logistic function, and the fitted R0 was obtained. 2 The value reached 0.997. Subsequently, the fitted curve was differentiated to obtain the corn grain filling rate (GFR) curve. Figure 3 c). The HGW growth curve and GFR curve show that the entire grain-filling process of maize is divided into a gradual increase period, a rapid increase period, and a slow increase period. The grain-filling rate is relatively slow during the gradual increase period, and accelerates during the rapid increase period, reaching its maximum about 20 days after flowering, with GFR reaching 1.249 g 100-grain. -1 d -1 During the slow-increase period, the grouting rate slows down again until it drops to 0.

[0085] Based on the Logistic curve of 100-grain weight growth, the weight gain of grains on days 10, 20, 30, and 40 after flowering was calculated, and correlation analysis was performed with Pn, Gs, Ci, and Tr. The results are as follows: Figure 4The correlation coefficients of Pn, Gs and Tr were high and mostly positive, with the highest correlation coefficients of 0.717, 0.658 and 0.654, respectively. The correlation of Ci was generally low, with the highest correlation coefficient of only 0.513. At 10, 20, 30 and 40 days after flowering, the maximum correlation coefficients of Pn and HGW occurred at 14:00, 18:00, 10:00 and 12:00, respectively. The maximum correlation coefficients of photosynthetic parameters and HGW at different grain-filling stages occurred at different times of the day, which increased the difficulty of evaluating photosynthetic capacity. The weight gain of the grain at 10, 20, 30 and 40 days after flowering was correlated with Pn_area, Gs_area, Ci_area and Tr_area. The correlation of Pn_area and Tr_area was significantly improved compared with Pn and Tr at a single time. The correlation coefficients of Pn_area at 10, 20, 30 and 40 days after flowering were 0.759, 0.813, 0.822 and 0.735, respectively, with the highest increase of 0.105 compared with Pn. The correlation coefficients of Tr_area at 10, 20, 30 and 40 days after flowering were 0.668, 0.660, 0.633 and 0.477, respectively, with the highest increase of 0.109 compared with Tr. The correlation of Gs_area and Ci_area with HGW was low, with the highest correlation coefficient of only 0.510, and the correlation did not improve or even decreased compared with Pn and Tr at a single time. The correlation of ODP with HGW was analyzed, and the results showed that the correlation coefficients of ODP at 10, 20, 30 and 40 days after flowering were 0.831, 0.882, 0.856 and 0.833, respectively. The r was significantly improved compared with Pn_area, Gs_area, Ci_area and Tr_area, indicating that the constructed ODP index could better represent the ability of maize photosynthesis to generate organic matter.

[0086] The sensitive wavebands were screened by two-dimensional correlation spectroscopy, and ODP was estimated by optimized vegetation index Figure 5 The results showed that the spectral response was different at different grain-filling stages. The most sensitive wavebands in the blue, green, red, red edge and near-infrared regions were 450 nm, 554 nm, 680 nm, 744 nm and 934 nm at 10 days after flowering, respectively. The most sensitive wavebands were 442 nm, 545 nm, 676 nm, 735 nm and 935 nm at 20 days after flowering, respectively. The most sensitive wavebands were 450 nm, 560 nm, 621 nm, 742 nm and 781 nm at 30 days after flowering, respectively. The most sensitive wavebands were 450 nm, 504 nm, 678 nm, 681 nm and 936 nm at 40 days after flowering, respectively.

[0087] Based on the results of two-dimensional correlation spectroscopy analysis, 17 VIs were optimized, and then correlated with ODP, as shown in Figure 6 At 10, 20, 30 and 40 days after flowering, the correlation of SIPI, VARI, SIPI and CARI was the highest, and r was-0.824, 0.886, -0.808 and 0.852, respectively. When the data of the whole grain filling period was comprehensively analyzed, BRI had the highest correlation among all VIs (r=0.732). Overall, the data of a single grain filling stage showed stronger correlation than the comprehensive data.

[0088] The data and model were evaluated using leave-one-out cross-validation Figure 7 ). The results showed that the inversion models constructed with all VIs and VIs after feature screening as input could achieve high prediction accuracy. Among them, the inversion model after CARS feature screening had the highest accuracy, R 2 , RMSE and MAE were 0.82, 111.03 and 89.14, respectively, and the inversion model without feature screening had the second highest accuracy, R 2 also reached 0.81, and RMSE and MAE were 113.19 and 92.14, respectively.

[0089] The test results show that the population photosynthetic capacity index proposed in the embodiment of the application is close to the actual situation, and can reflect the real field corn growth situation. In addition, the index can be quickly and accurately estimated by using a UAV, and the photosynthetic capacity identification efficiency is improved.

[0090] It can be known from the technical common sense that the application can be realized by other embodiments without departing from the spirit or essential characteristics thereof. Therefore, the above disclosed embodiments are only examples, and are not the only ones. All changes within the scope of the application or within the scope equivalent to the application are included in the application.

Claims

1. A method for constructing and estimating an index of photosynthetic capacity of a maize population, characterized by, The method comprises the following steps: S1: obtaining four photosynthetic parameters of corn net photosynthetic rate Pn, stomatal conductance Gs, intercellular CO2 concentration Ci and transpiration rate Tr, and measuring once every two hours from 6 am to 6 pm; S2: shortening the measurement interval of the net photosynthetic rate Pn, stomatal conductance Gs, intercellular CO2 concentration Ci and transpiration rate Tr to one hour by using Akima spline curve interpolation method; S3: fitting the values of the net photosynthetic rate Pn, stomatal conductance Gs, intercellular CO2 concentration Ci and transpiration rate Tr by using a sine and cosine function to obtain a fitting curve of the four parameters, and the sine and cosine function formula is: Wherein, a is the amplitude, b is the frequency, c is the phase constant of each sine wave, and n is the number of items in the sequence, 1≤n≤8; S4: Calculate the projected area under the curve using the integral method, get , , and ; S5: normalizing the projection area, and determining the weight of each projection area by using a CRITIC weight method; S6: calculating the photosynthesis index ODP according to the projection area and weight of each parameter, and the ODP calculation formula is: wherein, , , and are the weights of , , and respectively; S7: obtaining the hundred-grain weight of the seeds, simulating the corn filling process, and measuring the feasibility of the ODP index by the increase of the hundred-grain weight; S8: obtaining the hyperspectral image of the target region; S9: preprocessing the hyperspectral image of the target region, and converting the grid image of the target region into a hyperspectral reflectivity; S10: calculating 17 vegetation indices, screening sensitive bands by using two-dimensional correlation spectral analysis technology, and optimizing and improving the vegetation indices; S11: screening the vegetation indices by using a competitive adaptive reweighted sampling method, and removing the multicollinearity; S12: constructing an ODP estimation model based on the vegetation indices; S13: The coefficient of determination R 2 , the root mean square error RMSE and the mean absolute error MAE are used to evaluate the estimation performance of the model, the R 2 , RMSE, MAE calculation formula is: wherein, is the number of samples, and are the measured and estimated ODPs, respectively, is the average of the measured ODPs.

2. The method according to claim 1, wherein the method is characterized by: In step S1, natural light is used as the light source, three corn plants with consistent growth conditions, no pests and no damage are randomly selected, and the ear leaves are selected for measurement. The leaf blades are kept in the original position during the measurement process, and the average value is calculated and used for subsequent data analysis.

3. The method according to claim 1, wherein the method is characterized by: In step S5, the projection area is normalized, and the normalized interval is [0, 1], and the CRITIC weight method formula is: wherein, is the intra-index variability, is the inter-index conflictivity, is the index mean, is the correlation coefficient between indices i and j, denotes the information content, is the maximum weight.

4. The method according to claim 1, wherein the method is characterized by: In step S7, 30 plants with the same growth vigor are selected for marking at the silking stage of corn, 5 corn ears are selected at 10, 20, 30 and 40 days after flowering, respectively, 100 seeds are taken from the middle of each ear, the seeds are dried in an oven at 105°C for 30 minutes, and then further dried at 80°C until the weight is constant, the weight of 500 seeds is measured by an electronic scale, the hundred-grain weight HGW is calculated, the filling process of the seeds is simulated by using a Logistic equation, the number of days after flowering is used as the independent variable, and the HGW measured once every 10 days is used as the dependent variable, and the Logistic equation is: Wherein, t is the number of days after flowering, A is the final hundred-grain weight, B is the initial value parameter, and C is the growth rate parameter.

5. The method according to claim 1, wherein the method is characterized by: In step S8, the flight height of the UAV is 30 m, the heading overlap ratio and the lateral overlap ratio are 70% and 60% respectively, the reflectivity calibration is performed using a standard whiteboard before the UAV takes off, the atmospheric calibration is performed using gray cloths with reflectivities of 20%, 50% and 70% after the UAV takes off, the image acquisition is performed in the automatic exposure mode, and the hyperspectral image acquisition time is 11:00 on the day of the photosynthetic parameter measurement.

6. The method according to claim 1, wherein the method is characterized by: In step S9, after the image acquisition, the lens calibration, the reflectivity calibration and the atmospheric calibration are performed on the image, the calibrated image is spliced using the position and attitude measurement system, the latitude and longitude data and the feature point matching of the image overlap area, the hyperspectral raster image containing 176 bands is obtained, finally, the canopy reflectivity of each plot is extracted, the Savitzky-Golay method is used to perform the smoothing processing on the reflectivity curve, and the window size is set to 5.

7. The method for constructing and estimating maize population photosynthetic capacity index according to claim 1, characterized in that: In step S11, the sampling number is set to 100000, and the wavelength variable subset corresponding to the minimum value of the root mean square error is selected as the characteristic wavelength.

8. The method according to claim 1, wherein the method is characterized by: In step S12, the gradient method is used to divide the training set and the test set in a ratio of 3:1, and the specific operation is as follows: first, the samples are arranged from small to large according to the ODP value, then the second or third sample in every four samples is selected as the test set, and the remaining samples are the training set, the ODP value of the test set samples is ensured to be within the ODP value range of the training set samples and to be uniformly distributed, and the leave-one-out cross-validation is used to further evaluate the estimation performance of the model.

Images