A multispectral diagnosis method for nitrogen nutrition of rice and wheat for precision fertilization using drones

Abstract

The present invention discloses a rice and wheat nitrogen nutrition multispectral diagnosis method for precision fertilization by unmanned aerial vehicles, comprising: step 1, obtaining multispectral unmanned aerial vehicle remote sensing images and collecting rice and wheat agronomic parameters, and establishing a nitrogen nutrition spectral diagnosis model; step 2, calculating the critical nitrogen concentration of the aboveground part of rice and wheat based on the critical nitrogen concentration dilution model of dry matter, and further obtaining the critical nitrogen absorption amount; calculating the actual nitrogen absorption amount of the aboveground part of rice and wheat based on the nitrogen nutrition spectral diagnosis model; obtaining the recommended nitrogen topdressing amount of rice and wheat in the corresponding period based on the critical nitrogen absorption amount and the actual nitrogen absorption amount. The present invention acquires images of rice and wheat growth by multispectral unmanned aerial vehicles, establishes a dynamic diagnosis model of nitrogen nutrition parameters, realizes rapid diagnosis of nitrogen nutrition according to the critical nitrogen concentration dilution model and generates a fertilization prescription map, guides agricultural unmanned aerial vehicles to accurately fertilize according to the calculated recommended nitrogen application amount, reduces the input of fertilizers and manpower under the premise of ensuring stable yield of rice and wheat, and has good application prospects.

Description

Technical Field

[0001] The present invention relates to the field of modern agriculture, and in particular to remote sensing diagnosis of nitrogen nutrition of rice and wheat based on multispectral unmanned aerial vehicle images and a precision fertilization method of agricultural unmanned aerial vehicles based on critical nitrogen concentration. Background Art

[0002] The amount of nitrogen fertilizer applied significantly affects biomass accumulation and grain yield formation. In addition, the amount of nitrogen fertilizer applied affects the protein content of the grain, which in turn affects the quality of the grain. The yield of rice and wheat generally increases with the increase in the amount of nitrogen applied, but excessive use will not significantly increase the yield of crops, but will significantly reduce the utilization rate of nitrogen fertilizer, resulting in diminishing returns. At present, drone topdressing generally relies on experience, and most of them only focus on the uniformity of fertilizer application, pay less attention to the actual fertilizer needs of crops, and do not consider the relationship between fertilizer application and crop nutritional needs, resulting in low fertilizer utilization and the risk of nutrient loss.

[0003] Therefore, how to quickly and accurately diagnose nitrogen nutrition of plants, build a remote sensing diagnosis method for crop nitrogen nutrition profit and deficit suitable for precision fertilization by agricultural drones, and solve the mismatch between fertilization and crop nutrition needs are issues that need to be addressed urgently. Summary of the invention

[0004] The purpose of the present invention is to overcome the problem that in the existing technology, it is impossible to realize real-time nutritional diagnosis of crops to fertilize crops, but empirical fertilization is used and individual fertilization cannot be carried out for a certain area with poor growth. The present invention is based on multi-spectral drones to perform nutritional diagnosis of crops, use critical nitrogen concentration to calculate the nitrogen deficiency of plants in different growth periods, and use the variable fertilization system of agricultural drones to accurately fertilize plants at different times. Through this method, we can simply and quickly obtain the nutritional profit and loss status of plants, so that point-to-point agricultural drones can achieve precise fertilization.

[0005] To achieve the above object, the technical solution adopted by the present invention is: a multi-spectral diagnosis method of nitrogen nutrition of rice and wheat for precision fertilization by unmanned aerial vehicle, comprising the following steps:

[0006] Step 1, the rice and wheat nitrogen nutrition remote sensing diagnosis module obtains multispectral UAV remote sensing images and collects rice and wheat agronomic parameters, and uses a machine learning algorithm to establish a diagnostic model based on the multispectral UAV remote sensing images and the obtained nitrogen concentration.

[0007] Step 2, the topdressing decision model generates a plate. According to previous studies, the nitrogen concentration of crops decreases with the increase of dry matter accumulation, and the critical nitrogen concentration is the minimum nitrogen concentration required for the maximum dry matter accumulation amount obtained by crops in a certain growth period. The critical nitrogen concentration dilution model of dry matter is a model formed based on this theory, that is, Nc=a*AGB^(-b). The aboveground dry biomass of rice and wheat in the studied area is used as the horizontal coordinate, and the aboveground nitrogen concentration is used as the horizontal coordinate for modeling, and the critical nitrogen concentration dilution model of the area can be obtained. According to the critical nitrogen concentration dilution model based on dry matter mass proposed by predecessors, the aboveground (rice and wheat stems and leaves) critical nitrogen concentration of rice and wheat is calculated, and the model trained in step 1 is used to calculate the actual nitrogen absorption of the aboveground part of rice and wheat. The critical nitrogen absorption and the actual nitrogen absorption are calculated based on the critical nitrogen concentration of the aboveground part to calculate the recommended nitrogen absorption of rice and wheat in the corresponding period.

[0008] Step 3, the generation of the digital multi-dimensional prescription map and the agricultural drone precision fertilization section, using software such as SuperMap and GlobalMappper to save the recommended nitrogen application amount as an elevation image, generate a digital fertilization prescription map, and provide a "smart brain" for the precision operation of agricultural drones to guide agricultural drones to perform precise fertilization operations.

[0009] Preferably, step 1 comprises the following steps:

[0010] Step 1.1, obtain data, obtain multispectral drone images of rice and wheat under different nitrogen fertilizer gradients during the key fertilization period of rice and wheat (jointing stage and booting stage), and use the five-point sampling method to conduct destructive sampling in each plot.

[0011] Step 1.2, obtain agronomic parameters, and fix the collected rice and wheat samples at 105°C for 30 minutes, and then dry them at 80°C to constant weight. Weigh the dried samples, calculate the biomass, and then crush them. Use a Kjeldahl nitrogen analyzer to determine the plant nitrogen content (PNC) of the samples collected from each plot.

[0012] Step 1.3, multispectral UAV image processing, the acquired multispectral UAV images are stitched and radiometrically calibrated using DJI Zhitu software, and the digital quantization value (DN) is converted into reflectivity through reflectors with three reflectivities of 0.25, 0.5, and 0.95 to generate a TIFF file of the orthophoto of the study area.

[0013] Step 1.4, remove the soil background value. Since the plants are small in the early stage, the soil and water reflection between the plants have a large impact on the remote sensing diagnosis result image. Therefore, the soil and water background values ​​are removed for the multispectral drone images in the tillering and greening stages. NDVI is the normalized vegetation index, which is one of the most commonly used vegetation indices to reflect the growth of crops. Therefore, NDVI is used as the main vegetation index for soil, water and crop classification. Import the NDVI and true color remote sensing images directly calculated by DJI Zhitu into Arcmap software, find the critical NDVI value of soil and plants in the layer properties, enter “> critical NDVI value” in Arcmap’s “Raster Calculator”, set the generated raster name and storage path in “Output raster”, and you can generate new raster data with soil removed. Enter the generated new raster data in “Raster to polygon” to convert it into vector data. Use vector data to crop the original data to obtain an image with soil background removed.

[0014] Step 1.5, extract the vegetation index from the drone. Use ENVI's "band math" and the vegetation index formula to perform band operations on the multispectral drone image to calculate the vegetation index. Use Arcmap software to divide the multispectral drone image into regions of interest, and then extract the mean of the vegetation index in each region of interest, and use the mean to represent the vegetation index of each plot.

[0015] Step 1.6, data analysis, correlation analysis was performed on the vegetation index obtained in step 1.5 and the plant nitrogen content obtained in step 1.2: the vegetation index with the highest correlation with the plant nitrogen content was selected as the optimal vegetation index for linear modeling, and the optimal vegetation index was used as the horizontal axis and the plant nitrogen content was used as the vertical axis to perform linear, exponential, polynomial and logarithmic modeling. The absolute coefficient (R 2 ) and root mean square error (RMSE) were used as evaluation indicators to evaluate the model, and the optimal nitrogen nutrition spectral diagnostic model was selected and established.

[0016] Preferably, step 2 comprises the following steps:

[0017] Step 2.1, calculating the critical nitrogen concentration of rice and wheat based on the dry biomass obtained in step 1.2 and the critical nitrogen concentration dilution curve proposed by previous researchers;

[0018] Step 2.2, obtaining the aboveground nitrogen concentration y1 of the plant based on the nitrogen nutrition spectrum diagnostic model, and obtaining the actual nitrogen absorption of rice and wheat according to the product of the aboveground nitrogen concentration y1 of the plant and the dry biomass y2 of the plant;

[0019] Step 2.3, calculating the critical nitrogen uptake of rice and wheat according to the product of the critical nitrogen concentration of the plant calculated in step 2.1 and the dry biomass of the plant;

[0020] Step 2.4, calculating the difference between the critical nitrogen uptake of rice and the actual nitrogen uptake according to the difference between the actual nitrogen uptake of rice and the critical nitrogen uptake of rice calculated in step 2.2 and step 2.3;

[0021] Step 2.5, calculate the theoretical nitrogen deficiency, that is, the theoretical nitrogen application amount, based on the ratio of the current growth period of rice and wheat to the total growth days.

[0022] Preferably, in step 2.2, the dry biomass of the sample collected in step 1.2 is obtained and the remote sensing image is also obtained at the same time, the dry biomass is used as the vertical coordinate y, the vegetation index calculated using the remote sensing image and the dry biomass are subjected to correlation analysis, the vegetation index with the highest correlation is selected, and it is used as the horizontal coordinate x for modeling, and a model of y2=kx+b can be obtained, which is a model for monitoring the dry biomass of plants in this field through remote sensing images; the actual nitrogen absorption of rice and wheat is the product of y1 and y2.

[0023] Preferably, step 3 comprises the following steps:

[0024] Step 3.1, create a new data source in the SuperMap software and import the multispectral drone image tif file;

[0025] Step 3.2, create a surface data set in the SuperMap software according to the theoretical topdressing nitrogen amount (Nd) calculated in the above step, calculate the required fertilizer amount according to the nitrogen content of urea, and draw the data for the fields that need to be fertilized;

[0026] Step 3.3, convert the vector surface dataset into a shp dataset;

[0027] Step 3.4, import shp data in Arcmap and convert the input shp file into a raster;

[0028] Step 3.5, write the raster converted in step 3.4 into the raster data name with the extension name, change the pixel type to 3_BIT_FLOAT, and change the band to 1;

[0029] Step 3.6, convert the new raster embedded in step 3.5 into tif format;

[0030] Step 3.7, import the tif file in GlobalMappper, select 32-bit floating point samples, cancel the option of always generating square pixels, and generate a TFW file;

[0031] Step 3.8, save the elevation image according to the path saved by the agricultural drone, and generate a digital fertilization prescription map that can be recognized by the agricultural drone.

[0032] The present invention also discloses a system for multispectral diagnosis of nitrogen nutrition of rice and wheat for precision fertilization by unmanned aerial vehicles, including: a remote sensing diagnosis module for nitrogen nutrition of rice and wheat, which is used to obtain multispectral remote sensing images of unmanned aerial vehicles, collect agronomic parameters of rice and wheat, and establish a nitrogen nutrition spectral diagnosis model by combining the multispectral remote sensing images of unmanned aerial vehicles and the agronomic parameters obtained by using a machine learning algorithm;

[0033] The topdressing decision model generation module calculates the critical nitrogen concentration of the aboveground part of rice and wheat based on the critical nitrogen concentration dilution model of dry matter, and further calculates the critical nitrogen absorption amount; calculates the actual nitrogen absorption amount of the aboveground part of rice and wheat based on the rice and wheat nitrogen nutrition remote sensing diagnosis module; calculates the recommended topdressing amount of rice and wheat in the corresponding period based on the critical nitrogen absorption amount and the actual nitrogen absorption amount;

[0034] The generation of digital multi-dimensional prescription maps and the agricultural drone precision fertilization module use HyperMap and GlobalMapper software to save the recommended nitrogen application rate as an elevation image, generate a digital fertilization prescription map, and guide agricultural drones to perform precision fertilization operations. Beneficial Effects

[0035] The present invention uses a consumer-grade multispectral drone to acquire images of rice and wheat growth, establishes a dynamic diagnosis model for nitrogen nutrition parameters, realizes rapid diagnosis of nitrogen nutrition and generates a fertilization prescription map based on a critical nitrogen concentration dilution model, and guides agricultural drones to accurately fertilize based on the calculated recommended nitrogen application amount. This method reduces the input of fertilizers and manpower while ensuring stable rice and wheat production, is convenient to implement, and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is an implementation structure diagram of a rice and wheat nitrogen nutrition multi-spectral diagnosis method for drone precision fertilization of the present invention;

[0037] Figure 2 It is a system block diagram of the rice and wheat nitrogen nutrition remote sensing diagnosis module of the present invention;

[0038] Figure 3 This is a system block diagram of the rice and wheat topdressing decision model generation section in the present invention.

[0039] Figure 4 It is a system block diagram of the generation of digital multi-dimensional prescription map and the precision fertilization section of agricultural drone in the present invention;

[0040] Figure 5 This is a diagram of a vegetation area division experiment according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described below in conjunction with the accompanying drawings.

[0042] The present invention discloses a rice and wheat nitrogen nutrition multispectral diagnosis method for precision fertilization by unmanned aerial vehicles. The method includes three sections: a rice and wheat nitrogen nutrition remote sensing diagnosis section, a topdressing decision model generation section, a digital multidimensional prescription map generation section, and an agricultural unmanned aerial vehicle precision fertilization section. The present invention acquires images of rice and wheat growth by multispectral unmanned aerial vehicles, establishes a nitrogen nutrition parameter remote sensing diagnosis model, and establishes a file recognizable by agricultural unmanned aerial vehicles based on the recommended nitrogen application amount calculated according to the critical nitrogen concentration, so as to guide the precision fertilization operation of agricultural unmanned aerial vehicles. The method reduces the input of fertilizers and manpower while ensuring the stable yield of rice and wheat, is easy to use, and has good application prospects.

[0043] like Figures 1 to 4 As shown, the following steps are included:

[0044] Step 1, the rice and wheat nitrogen nutrition remote sensing diagnosis module obtains multispectral UAV remote sensing images and collects rice and wheat agronomic parameters, and uses a machine learning algorithm to establish a remote sensing diagnosis model based on the multispectral UAV remote sensing images and the obtained nitrogen concentration.

[0045] Step 1.1, obtain data, obtain multispectral drone images of rice and wheat under different nitrogen fertilizer gradients during the key fertilization period of rice and wheat, and use the five-point sampling method to perform destructive sampling on each plot. It should be noted that the "key fertilization period" is the jointing stage and the booting stage.

[0046] Step 1.2, obtain agronomic parameters, and fix the collected rice and wheat samples at 105°C for 30 minutes, and then dry them at 80°C to constant weight. Weigh the dried samples, calculate the biomass, and then crush them. Use a Kjeldahl nitrogen analyzer to determine the plant nitrogen content (PNC) of the samples collected from each plot.

[0047] Step 1.3, multispectral UAV image processing, the acquired multispectral UAV images are stitched and radiometrically calibrated using DJI Zhitu software, and the digital quantization value (DN) is converted into reflectivity through reflectors with three reflectivities of 0.25, 0.5, and 0.95 to generate a TIFF file of the orthophoto of the study area.

[0048] Step 1.4, remove the soil background value. Since the plants are small in the early stage, the soil and water reflection between the plants have a large impact on the diagnostic image. Therefore, the soil and water background values ​​are removed for the multispectral drone images in the tillering and greening stages. NDVI is the normalized vegetation index, which is one of the most commonly used vegetation indices to reflect the growth of crops. Therefore, NDVI is used as the main vegetation index for soil, water and crop classification. Import the NDVI and true color remote sensing images directly calculated by DJI Zhitu into Arcmap software, find the critical NDVI value of soil and plants in the layer properties, enter ">critical NDVI value" in "RasterCalculator" in Arcmap, set the generated raster name and storage path in "Output raster", and you can generate new raster data with soil removed. Enter the generated new raster data in "Raster to polygon" to convert it into vector data. Use vector data to crop the original data to obtain an image with soil background removed.

[0049] Step 1.5, extract the vegetation index from the drone. Use ENVI's "band math" and the vegetation index formula to perform band operations on the multispectral drone image to calculate the vegetation index. Use Arcmap software to divide the multispectral drone image into regions of interest, and then extract the mean of the vegetation index in each region of interest. Use the mean to represent the vegetation index of each plot, such as Figure 5 shown.

[0050] Step 1.6, data analysis, correlation analysis between the vegetation index obtained in step 1.5 and the plant nitrogen content obtained in step 1.2: the nitrogen content of the plant is calculated by the Kjeldahl method using a small amount of samples taken as above, and the vegetation index with the highest correlation with the plant nitrogen content is selected (the vegetation index is the value calculated from the image obtained by the drone) as the optimal vegetation index for linear modeling. The optimal vegetation index is used as the horizontal coordinate and the plant nitrogen content is used as the vertical coordinate to perform four basic modeling methods: linear, exponential, polynomial and logarithmic modeling. The absolute coefficient (R 2 ) and root mean square error (RMSE) were used as evaluation indicators to evaluate the model, and the vegetation index and nitrogen nutrient concentration were regressed and selected based on the machine learning algorithm to establish the optimal nitrogen nutrient spectral diagnostic model (i.e. Figure 2UAV remote sensing LNC monitoring model), expressed as y1=kx+b, which is a model for monitoring the nitrogen content of plants in this field through remote sensing images. In this embodiment, taking the jointing stage of wheat as an example: the remote sensing image is preprocessed and the spectrum is extracted to calculate the vegetation index, and the vegetation index with the highest correlation with the nitrogen content of the plant is selected, which is RVI, and the correlation coefficient r=-0.792. RVI is used as x to model the nitrogen content of the plant, and the model obtained is y=-0.8592x+5.2054 (x: is RVI, y is the nitrogen content of the plant), R 2 =0.58.

[0051] Step 2, calculate the critical nitrogen concentration of the aboveground part of rice and wheat according to the critical nitrogen concentration dilution model based on dry matter, use the model trained in step 1 to calculate the actual nitrogen absorption of the aboveground part of rice and wheat, calculate the critical nitrogen absorption based on the critical nitrogen concentration of the aboveground part, and calculate the recommended nitrogen topdressing amount of rice and wheat in the corresponding period based on the critical nitrogen absorption and the actual nitrogen absorption. It should be noted that "aerial part" refers to the aboveground part of the plant, which refers to the stems, leaves, flowers, fruits, and seeds of the plant. In this application, the aboveground part of rice and wheat refers to stems and leaves.

[0052] Step 2.1, calculate the critical nitrogen concentration of rice and wheat based on the dry biomass obtained in step 1.2 and the critical nitrogen concentration dilution curve proposed by previous researchers.

[0053] The nitrogen concentration of crops decreases with the increase of dry matter accumulation. The critical nitrogen concentration is the minimum nitrogen concentration required for the maximum dry matter accumulation of crops in a certain growth period. The critical nitrogen concentration dilution model of dry matter is a model formed based on this theory, expressed as Nc=a*AGB^(-b), where Nc is the critical nitrogen concentration of the aboveground part of the plant, AGB is the dry biomass of the aboveground part of the plant, and a is the aboveground dry biomass of 1 t·hm –2 The nitrogen concentration at that time, b represents the ratio of the relative accumulation rate of nitrogen to the biomass. Step 2.2, the actual nitrogen absorption of rice and wheat is obtained according to the product of the aboveground nitrogen concentration of the plant and the dry biomass of the plant. In step 1.2, the dry biomass of the collected samples was obtained and the remote sensing image was also obtained. The dry biomass was used as the ordinate y, and the vegetation index calculated using the remote sensing image was correlated with the dry biomass. The vegetation index with the highest correlation was selected and used as the abscissa x for modeling, and the model of y2=kx+b was obtained. This model is the model for monitoring the dry biomass of plants in this field through remote sensing images. The actual nitrogen absorption of rice and wheat is the product of y1 and y2. Step 2.3, the critical nitrogen absorption of rice and wheat is calculated according to the product of the critical nitrogen concentration of the plant and the dry biomass of the plant calculated in step 2.1.

[0054] Step 2.4, calculating the difference between the critical nitrogen uptake of rice and the actual nitrogen uptake according to the difference between the actual nitrogen uptake of rice and the critical nitrogen uptake of rice calculated in steps 2.2 and 2.3.

[0055] Step 2.5, calculate the theoretical nitrogen deficiency, that is, the theoretical nitrogen application amount, based on the ratio of the current growth period of rice and wheat to the total growth days.

[0056] Step 3, the generation of the digital multi-dimensional prescription map and the agricultural drone precision fertilization section, using software such as SuperMap and GlobalMappper to save the recommended nitrogen application amount as an elevation image, generate a digital fertilization prescription map, and provide a "smart brain" for the precise operation of agricultural drones to guide agricultural drones to perform precise fertilization operations.

[0057] Step 3.1, create a new data source in the SuperMap software (i.e., create a new blank page in the software) and import the multispectral drone image tif file (i.e., the TIFF file of the orthophoto of the study area in step 1.3).

[0058] Step 3.2, create a surface data set in the Hypergraph software based on the theoretical topdressing nitrogen amount (Nd) calculated in the above step, calculate the amount of fertilizer required based on the urea nitrogen content, and draw the data for the fields that need to be fertilized. The urea nitrogen content is generally between 42% and 46%, which is based on the urea nitrogen content of the specific fertilizer used.

[0059] Step 3.3, convert the vector surface dataset into a shp dataset.

[0060] Step 3.4, import the shp data in Arcmap and convert the input shp file into a raster.

[0061] Step 3.5, write the raster converted in step 3.4 into a custom raster data name with an extension, change the pixel type to 3_BIT_FLOAT, and change the band to 1.

[0062] Step 3.6, convert the new raster embedded in step 3.5 into tif format.

[0063] Step 3.7, import the tif file in GlobalMapper, select 32-bit floating point samples, uncheck Always generate square pixels, and generate a TFW file.

[0064] Step 3.8, save the elevation image according to the path saved by the agricultural drone, and generate a digital fertilization prescription map that can be recognized by the agricultural drone.

[0065] The present invention utilizes remote sensing technology to carry out real-time monitoring and nutrition diagnosis of farmland crops, and can fertilize according to the fields and seedlings, thereby reducing the error caused by human experience in fertilizing and the problem that field fertilizer cannot be supplied on demand.

[0066] Finally, it should be noted that the above is only a preferred implementation case of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multispectral diagnosis method for nitrogen nutrition of rice and wheat for precision fertilization by unmanned aerial vehicles, characterized in that: The following steps are involved: Step 1, obtaining multispectral UAV remote sensing images and collecting rice and wheat agronomic parameters, and establishing a nitrogen nutrition spectral diagnosis model based on the remote sensing images and rice and wheat agronomic parameters; Step 1.1, data acquisition: multispectral UAV images of rice and wheat under different nitrogen fertilizer gradients were acquired during the key fertilization period of rice and wheat; the multispectral UAV images were divided into several areas of interest, and destructive sampling was performed on each area using the five-point sampling method; Step 1.2, obtaining agronomic parameters: processing rice and wheat samples to obtain dry biomass and measuring the nitrogen content of the plants; Step 1.3, multispectral UAV image processing: stitch the acquired multispectral UAV images and perform radiometric calibration at the same time, convert the digital quantization values ​​into reflectivity based on a set of reflectors with different reflectivity, and generate a TIFF file of the orthophoto of the study area; Step 1.4, soil background removal: During the critical fertilization period of rice and wheat, soil and water background values ​​are removed from multispectral drone images at the tillering and greening stages; Step 1.5, extracting the vegetation index from drones: Use ENVI's "band math" and vegetation index formula to perform band operations on multispectral drone images and calculate the vegetation index; extract the mean of the vegetation index in each area of ​​interest, and use the mean to represent the vegetation index of each plot; Step 1.6, data analysis: perform correlation analysis on the vegetation index obtained in step 1.5 and the plant nitrogen content obtained in step 1.2 to establish a nitrogen nutrition spectrum diagnostic model; Step 2, based on the critical nitrogen concentration dilution model of dry matter, calculate the critical nitrogen concentration of the aboveground part of rice and wheat, and further obtain the critical nitrogen absorption; calculate the actual nitrogen absorption of the aboveground part of rice and wheat based on the nitrogen nutrition spectrum diagnosis model; based on the critical nitrogen absorption and the actual nitrogen absorption, obtain the recommended nitrogen topdressing amount of rice and wheat in the corresponding period, specifically, Step 2.1, calculating the critical nitrogen concentration of rice and wheat according to the dry biomass in step 1.2 combined with the critical nitrogen concentration dilution curve; Step 2.2, obtaining the aboveground nitrogen concentration y1 of the plant based on the nitrogen nutrition spectrum diagnostic model, and obtaining the actual nitrogen absorption of rice and wheat according to the product of the aboveground nitrogen concentration y1 of the plant and the dry biomass y2 of the plant; Step 2.3, calculating the critical nitrogen uptake of rice and wheat according to the product of the critical nitrogen concentration of the rice and wheat obtained in step 2.1 and the dry biomass; Step 2.4, calculating the difference between the actual nitrogen uptake of rice and wheat and the critical nitrogen uptake of rice and wheat calculated in steps 2.2 and 2.3; Step 2.5, calculate the theoretical nitrogen deficiency, i.e. the theoretical nitrogen application amount Nd, according to the ratio of the current growth period of rice and wheat to the total growth days; Step 3: Save the recommended nitrogen topdressing amount as an elevation image, generate a digital fertilization prescription map, and guide the agricultural drone to accurately fertilize. Specifically, Step 3.1, create a new data source in the SuperMap software and import the multispectral drone image tif file; Step 3.2, create a surface data set in the Hypergraph software according to the theoretical nitrogen application amount Nd, calculate the required fertilizer amount according to the nitrogen content of urea, and draw the field blocks that need to be fertilized to add data; Step 3.3, convert the vector surface dataset into a shp dataset; Step 3.4, import shp data in Arcmap and convert the input shp file into a raster; Step 3.5, write the raster converted in step 3.4 into the raster data name with the extension name, change the pixel type to 3_BIT_FLOAT, and change the band to 1; Step 3.6, convert the new raster embedded in step 3.5 into tif format; Step 3.7, import the tif file in GlobalMappper, select 32-bit floating point samples, cancel the option of always generating square pixels, and generate a TFW file; Step 3.8, save the elevation image according to the path saved by the agricultural drone, and generate a digital fertilization prescription map that can be recognized by the agricultural drone.

2. The multispectral diagnosis method for nitrogen nutrition of rice and wheat for precision fertilization by unmanned aerial vehicles according to claim 1, It is characterized in that The method for establishing a nitrogen nutrition spectral diagnostic model is specifically as follows: selecting the vegetation index with the highest correlation with the nitrogen content of the plant as the optimal vegetation index for linear modeling, using the optimal vegetation index as the horizontal coordinate x and the nitrogen content of the plant as the vertical coordinate y for linear, exponential, polynomial and logarithmic modeling, using the absolute coefficient R2 and the root mean square error RMSE as evaluation indicators of the model to evaluate the model, establishing the optimal nitrogen nutrition spectral diagnostic model, and obtaining the model of y1=kx+b, which is a model for monitoring the nitrogen content of the plants in this field through remote sensing images.

3. The multi-spectral diagnosis method for nitrogen nutrition of rice and wheat for precision fertilization by unmanned aerial vehicles according to claim 2, It is characterized in that The step 1.2 is specifically as follows: the collected rice and wheat samples are sterilized at 105°C±5°C for 30min±5min, and then dried at 80°C±5°C to constant weight; the dried samples are weighed, the biomass is calculated, and then they are crushed, and the plant nitrogen content PNC of the samples collected from each plot is determined using a nitrogen analyzer.

4. The rice and wheat nitrogen nutrition multispectral diagnosis method for drone precision fertilization according to claim 2, characterized in that: The specific steps of removing the background value in step 1.4 are as follows: Step 1.41, NDVI is the normalized difference vegetation index: NDVI is used as the main vegetation index for soil, water and crop classification to generate new raster data without soil; Step 1.42, import the NDVI and true color remote sensing images calculated directly by DJI Zhitu into Arcmap software, find the critical NDVI value of soil and plants in the layer properties, enter ">Critical NDVI value" in Arcmap's "Raster Calculator", set the generated raster name and storage path in "Output raster", and generate new raster data without soil; Step 1.43, input the generated new raster data into Arcmap's "Raster to polygon" to convert it into vector data; use the vector data to crop the original data to obtain an image without the soil background.

5. The multi-spectral diagnosis method for nitrogen nutrition of rice and wheat for precision fertilization by unmanned aerial vehicles according to claim 4, It is characterized in thatIn step 2.2, the dry biomass of the sample collected in step 1.2 is obtained and the remote sensing image is also obtained. The dry biomass is used as the vertical coordinate y, and the vegetation index calculated using the remote sensing image is correlated with the dry biomass. The vegetation index with the highest correlation is selected and used as the horizontal coordinate x for modeling, and the model y2=kx+b can be obtained. This model is a model for monitoring the dry biomass of plants in this field through remote sensing images; the actual nitrogen absorption of rice and wheat is the product of y1 and y2.

6. A system for multispectral diagnosis of nitrogen nutrition of rice and wheat for precision fertilization by unmanned aerial vehicles, characterized in that: base The rice and wheat nitrogen nutrition multispectral diagnosis method for drone precision fertilization according to claim 1 comprises: The rice and wheat nitrogen nutrition remote sensing diagnosis module is used to obtain multispectral UAV remote sensing images, collect rice and wheat agronomic parameters, and use machine learning algorithms to combine multispectral UAV remote sensing images with the obtained agronomic parameters to establish a nitrogen nutrition spectral diagnosis model; The topdressing decision model generation module calculates the critical nitrogen concentration of the aboveground part of rice and wheat based on the critical nitrogen concentration dilution model of dry matter, and further calculates the critical nitrogen absorption amount; calculates the actual nitrogen absorption amount of the aboveground part of rice and wheat based on the rice and wheat nitrogen nutrition remote sensing diagnosis module; calculates the recommended topdressing amount of rice and wheat in the corresponding period based on the critical nitrogen absorption amount and the actual nitrogen absorption amount; The generation of digital multi-dimensional prescription maps and the agricultural drone precision fertilization module use HyperMap and GlobalMapper software to save the recommended nitrogen application rate as an elevation image, generate a digital fertilization prescription map, and guide agricultural drones to perform precision fertilization operations.

Images