Fertilization prescription map rapid generation method and system based on multispectral detection

Abstract

The invention relates to a fertilization prescription map rapid generation method and system based on multispectral detection, and belongs to the technical field of agriculture. Comprising the following steps: multi-source data acquisition: synchronously acquiring a 450-900nm wave band crop multi-spectral image acquired by a multi-spectral camera carried by an unmanned aerial vehicle, laboratory detection nutrient data of a preset soil sampling point, real-time soil humidity data, plot boundary vector data and a crop recommended fertilization model; data preprocessing: carrying out radiation correction, geometric correction and de-noising processing on the multispectral image; performing crop nutrient demand inversion; according to the method, the multi-spectral image of the unmanned aerial vehicle, the soil sampling data, the real-time soil humidity and the plot boundary vector data are synchronously integrated, a raster data set of a unified coordinate system is constructed, and crop nutrient requirements and environment variables are comprehensively covered. In combination with vegetation index combinations such as NDVI, RENDVI and RVI, the crop nutrient content is inverted through the PLSR algorithm, and the prediction precision R2 is greater than 0.85 and is significantly superior to that of a single index model.

Description

Technical Field

[0001] This invention relates to the field of agricultural technology, and in particular to a method and system for rapidly generating fertilizer prescription maps based on multispectral detection. Background Technology

[0002] In modern agricultural production, fertilizer prescription maps are the core basis for variable fertilization in precision agriculture, and their accuracy and generation efficiency directly affect fertilizer utilization and crop yield. Currently, existing fertilizer prescription map generation technologies have the following limitations: Traditional soil sampling methods involve collecting soil samples from multiple locations in the field, testing the content of nutrients such as nitrogen, phosphorus, and potassium in the laboratory, and then creating a prescription map based on crop requirements. However, this method suffers from problems such as low sampling density (making it difficult to reflect the spatial heterogeneity of nutrients in the field), long testing cycles (usually 3-7 days), and high costs (sample testing costs account for up to 60%). Furthermore, it cannot reflect the dynamic nutrient requirements of crops during their growth period in real time.

[0003] Methods based on single spectral data: Crop vegetation indices (such as NDVI) are obtained using remote sensing or near-ground spectral equipment, and nutrient content is retrieved through empirical models. However, single spectral bands (such as the visible light band) are easily affected by soil background and crop growth status (such as leaf aging), resulting in low retrieval accuracy (nitrogen content retrieval error often exceeds 15%), and they do not take into account basic soil fertility, leading to a disconnect between prescription maps and actual needs.

[0004] The existing system has low generation efficiency: traditional methods require manual processing of spectral data, matching of soil parameters, and calculation of fertilizer application. The generation of prescription maps for a single plot (100 mu) takes more than 8 hours, which cannot meet the timeliness requirements of large-scale agricultural production, and is especially unsuitable for the topdressing needs of crops during their rapid growth period.

[0005] Existing technologies suffer from problems such as insufficient accuracy, slow generation, and high cost. There is an urgent need for a method and system that combines multispectral detection technology to generate fertilizer prescription maps quickly and accurately. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method and system for rapid generation of fertilizer prescription maps based on multispectral detection, which aims to solve the problems in the background art.

[0007] To achieve the aforementioned objectives, the first aspect of this invention proposes a method and system for rapidly generating fertilizer prescription maps based on multispectral detection, comprising the following steps: (1) Multi-source data acquisition: Simultaneously acquire 450-900nm band crop multispectral images collected by UAV equipped with multispectral camera, laboratory nutrient data of preset soil sampling points, real-time soil moisture data, plot boundary vector data and crop recommended fertilization model; (2) Data preprocessing: Radiometric correction, geometric correction and denoising processing are performed on the multispectral images, vegetation index combination is calculated, and the vegetation index data is spatially registered with soil basic data and plot boundary data to generate a raster dataset with a unified coordinate system. (3) Crop nutrient demand inversion: Using vegetation index combination as input, a crop nutrient inversion model is constructed based on partial least squares regression (PLSR) algorithm, outputting the current nutrient content of the crop in each grid cell, and calculating the nutrient deficit in combination with the crop growth period requirements; (4) Fertilizer application rate calculation: Combining the basic soil nutrient content and crop nutrient deficit, the fertilizer application rate per unit area is calculated according to the following formula: Q = (D - S * N) / F in: Q represents the amount of fertilizer applied per unit area. D represents the crop nutrient deficit. S represents the effective soil supply (determined by soil test values ​​and element conversion coefficients). N represents soil nutrient utilization rate. F represents fertilizer utilization rate; (5) Prescription map generation and optimization: The fertilizer amount is assigned to the corresponding grid to generate the initial prescription map, the mutation value is eliminated by the spatial smoothing algorithm, and it is converted into the prescription output of the agricultural machinery operation unit.

[0008] Optionally, step (4) further includes: By integrating real-time weather warning data, when the predicted rainfall for the next 24 hours exceeds 30mm, the fertilizer application rate will be automatically dynamically adjusted. Apply a water loss prevention coefficient of 0.7 to sloping areas (slope > 5°); For low-lying areas (soil moisture > 85%), apply a fertilizer damage avoidance coefficient of 0.8; The revised fertilizer application rate is updated synchronously in the prescription map generation module.

[0009] Optionally, the resolution of the multispectral image in step (1) is no higher than 0.5 meters, and the soil sampling points are arranged according to a preset density to detect the N, P2O5, and K2O content of the topsoil in the 0-20cm layer.

[0010] Optionally, the vegetation index combination includes NDVI, RENDVI, and RVI, and the PLSR inversion model meets the predetermined accuracy requirements.

[0011] Optionally, the element conversion coefficients in step (4) are 0.6 for nitrogen, 0.5 for phosphorus, and 0.7 for potassium; For areas with soil moisture >80%, a correction factor of 0.8-0.9 is introduced to reduce the amount of fertilizer applied.

[0012] Optionally, in step (5), a moving window averaging algorithm is used for spatial smoothing, and the raster data is converted into a work unit prescription according to the width of the agricultural machinery operation, with the output format being a shp file.

[0013] This invention also protects a rapid fertilizer prescription map generation system based on multispectral detection, characterized in that it includes: Data acquisition module: Acquires UAV multispectral imagery, soil testing data, real-time soil moisture data, and plot boundary vectors; Data processing module: performs image correction, vegetation index calculation, and spatial registration of multi-source data; Nutrient Inversion Module: Constructs a crop nutrient model based on the PLSR algorithm and outputs raster-level nutrient deficit; Fertilization decision module: integrates basic soil nutrients, real-time weather warning data and topographic data to calculate dynamically adjusted fertilization amounts; Prescription diagram generation module: Generates and outputs prescription diagrams for agricultural machinery that can be recognized by agricultural machinery.

[0014] Optionally, the data acquisition module integrates a meteorological API interface to acquire rainfall and temperature forecast data in real time.

[0015] Optionally, the fertilization decision module has a built-in element conversion coefficient library, a regional correction rule library, and a user-defined interface.

[0016] Optionally, the prescription map generation module includes a topology verification unit for detecting the matching degree between the agricultural machinery operation path and the plot boundary, and automatically triggering data resampling for abnormal areas.

[0017] The beneficial effects of this invention are: 1. This invention provides a method and system for rapid generation of fertilization prescription maps based on multispectral detection. By simultaneously integrating UAV multispectral imagery, soil sampling data, real-time soil moisture, and plot boundary vector data, a raster dataset with a unified coordinate system is constructed, comprehensively covering crop nutrient requirements and environmental variables. Combining vegetation indices such as NDVI, RENDVI, and RVI, the PLSR algorithm is used to invert crop nutrient content, achieving a prediction accuracy of R... 2 >0.85, significantly better than the single index model. Based on the effective soil supply (S=soil test value×element conversion coefficient) and crop nutrient deficit, the fertilizer application rate per unit area is accurately calculated using the formula Q=(DS×N) / F, saving 20%-50% of chemical fertilizers compared to traditional methods, and improving fertilization accuracy and resource utilization.

[0018] 2. This invention provides a method and system for rapidly generating fertilizer prescription maps based on multispectral detection. It integrates real-time rainfall and temperature data. When the predicted 24-hour rainfall exceeds 30mm, it applies a loss prevention coefficient of 0.7 to sloping areas and a fertilizer damage avoidance coefficient of 0.8 to low-lying areas, preventing fertilizer loss or crop damage caused by heavy rain. For areas with soil moisture >80%, it automatically introduces a correction coefficient of 0.8-0.9 to reduce the risk of decreased fertilizer absorption efficiency under excessively wet conditions and minimize resource waste.

[0019] 3. This invention provides a method and system for rapid generation of fertilizer prescription maps based on multispectral detection. It employs a moving window averaging algorithm to eliminate abrupt changes in the prescription map, aggregates raster data into shapefile (SHP) format work units according to the agricultural machinery operation width, and is directly compatible with mainstream agricultural machinery control systems, eliminating the need for manual secondary processing. The prescription map generation module has a built-in topology verification unit to detect the matching degree between the agricultural machinery path and the plot boundary, automatically triggering data resampling for abnormal areas to ensure operational continuity and prescription map reliability.

[0020] 4. The present invention provides a method and system for rapid generation of fertilizer prescription maps based on multispectral detection, which reduces fertilizer usage and lowers agricultural production costs through precise fertilization. According to the case of Yongcheng Yinong Cooperative, variable fertilization by drones can save 10%-20% of fertilizer while improving fertilizer utilization. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of a method for rapidly generating fertilizer prescription maps based on multispectral detection, provided by an exemplary embodiment of the present invention.

[0022] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0023] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0024] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection, a direct connection, or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0025] Example 1 Reference Figure 1An embodiment of the present invention provides a method for rapidly generating fertilizer prescription maps based on multispectral detection, comprising the following steps: (1) Multi-source data acquisition: Simultaneously acquire 450-900nm band crop multispectral images collected by UAV equipped with multispectral camera, laboratory nutrient data of preset soil sampling points, real-time soil moisture data, plot boundary vector data and crop recommended fertilization model; (2) Data preprocessing: Radiometric correction, geometric correction and denoising processing are performed on the multispectral images, vegetation index combination is calculated, and the vegetation index data is spatially registered with soil basic data and plot boundary data to generate a raster dataset with a unified coordinate system. (3) Crop nutrient demand inversion: Using vegetation index combination as input, a crop nutrient inversion model is constructed based on partial least squares regression (PLSR) algorithm, outputting the current nutrient content of the crop in each grid cell, and calculating the nutrient deficit in combination with the crop growth period requirements; (4) Fertilizer application rate calculation: Combining the basic soil nutrient content and crop nutrient deficit, the fertilizer application rate per unit area is calculated according to the following formula: Q = (D - S * N) / F in: Q represents the amount of fertilizer applied per unit area. D represents the crop nutrient deficit. S represents the effective soil supply (determined by soil test values ​​and element conversion coefficients). N represents soil nutrient utilization rate. F represents fertilizer utilization rate; (5) Prescription map generation and optimization: The fertilizer amount is assigned to the corresponding grid to generate the initial prescription map, the mutation value is eliminated by the spatial smoothing algorithm, and it is converted into the prescription output of the agricultural machinery operation unit.

[0026] It should be noted that the multispectral image acquisition was carried out by a drone equipped with a multispectral camera (covering the 450-900nm band) to acquire crop canopy images at a resolution of less than 0.5 meters. The camera uses a light sensor to correct reflectivity data in real time to ensure data consistency under different lighting conditions.

[0027] Soil data collection: Soil sampling points were arranged within the plot at a preset density (e.g., 5-8 sampling points per hectare), and the nitrogen (N), phosphorus (P2O5), and potassium (K2O) content of the 0-20cm topsoil layer was tested in the laboratory. At the same time, soil moisture content was monitored in real time using a soil moisture sensor.

[0028] Data integration: Multispectral imagery, soil testing data, plot boundary vector data, and crop fertilization recommendation models are imported into a unified data platform to form a multi-source dataset.

[0029] By simultaneously acquiring multispectral imagery, soil nutrient data, and plot boundary information, the limitations of a single data source are overcome, improving the comprehensiveness and accuracy of prescription maps. Image resolution below 0.5 meters captures subtle differences in crop canopy, providing high-precision foundational data for subsequent vegetation index calculations. Real-time soil moisture monitoring, combined with data from pre-set soil sampling points, ensures the timeliness and accuracy of soil nutrient supply calculations.

[0030] In some embodiments, step (4) further includes: By integrating real-time weather warning data, when the predicted rainfall for the next 24 hours exceeds 30mm, the fertilizer application rate will be automatically dynamically adjusted. Apply a water loss prevention coefficient of 0.7 to sloping areas (slope > 5°); For low-lying areas (soil moisture > 85%), apply a fertilizer damage avoidance coefficient of 0.8; The revised fertilizer application rate is updated synchronously in the prescription map generation module.

[0031] It should be noted that the meteorological data access system obtains real-time rainfall and temperature forecast data for the next 24 hours through a meteorological API interface.

[0032] Dynamic correction logic: When the predicted rainfall is >30mm, apply a loss prevention coefficient of 0.7 to areas with a slope >5° (identified by DEM data) to reduce the risk of fertilizer loss; apply a fertilizer damage avoidance coefficient of 0.8 to low-lying areas with soil moisture >85% to avoid crop damage caused by excessive fertilization.

[0033] Automatic correction: The corrected fertilization amount is updated to the prescription map generation module in real time to ensure that the prescription map is synchronized with environmental changes.

[0034] By dynamically adjusting fertilizer application rates through meteorological early warnings, fertilizer loss or damage caused by extreme weather (such as heavy rain) can be avoided, thus improving fertilization safety. Differentiated remediation strategies for sloping and low-lying areas significantly reduce the risk of soil erosion and compaction, meeting the needs of green agriculture.

[0035] In one example, the resolution of the multispectral image in step (1) is no higher than 0.5 meters, and the soil sampling points are laid out at a preset density to detect the N, P2O5, and K2O content of the topsoil layer of 0-20cm.

[0036] It should be noted that the image resolution is controlled by setting the multispectral camera to a resolution of less than 0.5 meters to ensure that the differences in the reflectance of the canopy of individual crops can be effectively identified.

[0037] Soil sampling point layout: Based on the plot area (e.g., 8-10 sampling points for a 100-mu plot), use grid-based or random sampling methods to cover different terrains and crop growth areas.

[0038] The combination of high-resolution imagery and a reasonable sampling point density ensures the spatial representativeness of soil nutrient data and crop growth data, reducing data bias. By scientifically deploying sampling points, balancing data accuracy and sampling costs, this method is suitable for promotion in small and medium-sized farmlands.

[0039] In one example, the vegetation index combination includes NDVI, RENDVI, and RVI, and the PLSR inversion model meets a predetermined accuracy requirement.

[0040] It should be noted that NDVI (Normalized Difference Vegetation Index), RENDVI (Red Edge Normalized Difference Vegetation Index), and RVI (Ratio Vegetation Index) are extracted from multispectral images, and their combined analysis enhances the sensitivity of crop nutrient retrieval.

[0041] The PLSR model is constructed using vegetation indices as input variables and employs the Partial Least Squares Regression (PLSR) algorithm to establish a crop nutrient inversion model. Model training requires satisfying R... 2 >0.85, RMSE <0.15, to ensure prediction accuracy.

[0042] The combination of NDVI, RENDVI, and RVI can cover the spectral characteristics of different crop growth stages, improving the model's ability to retrieve key nutrients such as nitrogen. The PLSR algorithm reduces redundant variable interference through dimensionality reduction, improving the model's stability in complex farmland environments.

[0043] In one example, the element conversion coefficients in step (4) are 0.6 for nitrogen, 0.5 for phosphorus, and 0.7 for potassium; For areas with soil moisture >80%, a correction factor of 0.8-0.9 is introduced to reduce the amount of fertilizer applied.

[0044] It should be noted that, based on the soil-crop nutrient transformation law, the nitrogen conversion coefficient is set to 0.6, phosphorus to 0.5, and potassium to 0.7, used to calculate the effective soil supply (S=N×conversion coefficient). When soil moisture > 80%, a correction coefficient of 0.8-0.9 is automatically introduced to reduce the amount of fertilizer applied to avoid a decrease in fertilizer absorption efficiency under excessively wet conditions.

[0045] The element conversion coefficient, based on soil science theory, ensures the rationality of soil nutrient supply calculations and reduces human error. The moisture correction rule dynamically adjusts fertilizer application rates, preventing fertilizer waste or crop damage caused by soil moisture saturation.

[0046] In one example, step (5) uses a moving window averaging algorithm for spatial smoothing and converts the raster data into a work unit prescription based on the width of the agricultural machinery operation, with the output format being a shp file.

[0047] It should be noted that the spatial smoothing process employs a moving window averaging algorithm (with a window size of 3×3 grids) to eliminate local abrupt changes in the prescription map, ensuring the continuity of fertilizer application distribution. Based on the agricultural machinery operating width (e.g., 2 meters), the grid data is aggregated into operating units and output in SHP file format, compatible with mainstream agricultural machinery control systems.

[0048] Example 2 An embodiment of the present invention provides a rapid fertilizer prescription map generation system based on multispectral detection, characterized in that it includes: Data acquisition module: Acquires UAV multispectral imagery, soil testing data, real-time soil moisture data, and plot boundary vectors; Data processing module: performs image correction, vegetation index calculation, and spatial registration of multi-source data; Nutrient Inversion Module: Constructs a crop nutrient model based on the PLSR algorithm and outputs raster-level nutrient deficit; Fertilization decision module: integrates basic soil nutrients, real-time weather warning data and topographic data to calculate dynamically adjusted fertilization amounts; Prescription diagram generation module: Generates and outputs prescription diagrams for agricultural machinery that can be recognized by agricultural machinery.

[0049] It should be noted that the data acquisition module integrates UAV multispectral image acquisition, soil sensor data uploading, and meteorological API interface, supporting real-time synchronization of multi-source data.

[0050] The data processing module performs image radiometric correction (using pseudo-standard ground feature radiometric correction method), geometric correction (registration error ≤ 0.5 pixels) and vegetation index calculation.

[0051] The nutrient inversion module outputs raster-level nutrient deficits based on the PLSR model and generates a dynamic fertilization demand table by combining the crop's growth stage requirements.

[0052] The fertilization decision module integrates soil data, weather warnings, and topographic information to calculate and dynamically adjust the amount of fertilizer to be applied.

[0053] The prescription map generation module generates prescription maps in shapefile format and uses a topology verification unit to detect the matching degree between the agricultural machinery operation path and the plot boundary. Abnormal areas are automatically triggered for data resampling.

[0054] Each functional module operates independently yet works collaboratively, facilitating system upgrades and maintenance and reducing development costs. The entire process, from data collection to prescription output, is automated, minimizing manual intervention and improving efficiency.

[0055] In some embodiments, the data acquisition module integrates a meteorological API interface to acquire rainfall and temperature forecast data in real time.

[0056] It should be noted that the data acquisition module connects to a meteorological platform (such as the China Meteorological Administration or a third-party meteorological service) via an API interface to obtain real-time rainfall, temperature, and wind speed forecast data, which is then transmitted to the fertilization decision module. Real-time updates of meteorological data ensure that the fertilization decision module can quickly respond to environmental changes, improving the dynamic adaptability of the prescription map.

[0057] In some embodiments, the fertilization decision module has a built-in element conversion coefficient library, a regional correction rule library, and a user-defined interface.

[0058] It should be noted that the element conversion coefficient library pre-stores the conversion coefficients and correction rules for nitrogen, phosphorus, and potassium, supporting parameter calls for different crops and soil types. The regional correction rule library includes slope correction, humidity correction, and fertilizer damage avoidance rules, and supports users to add new custom rules.

[0059] The rule base design allows the system to adapt to different crops, soil and climate conditions, and the user-defined interface further meets personalized needs.

[0060] In some embodiments, the prescription map generation module includes a topology verification unit for detecting the matching degree between the agricultural machinery operation path and the plot boundary, and automatically triggering data resampling for abnormal areas.

[0061] It should be noted that the topology verification unit uses GIS tools to detect the matching degree between the agricultural machinery operation path and the plot boundary in the prescription map, identifying disconnected or overlapping areas. For areas with a matching degree below the threshold, the anomaly handling mechanism triggers a data resampling process, regenerates the local prescription map, and updates the overall output.

[0062] Topology verification ensures that the agricultural machinery operation path matches the plot boundary perfectly, avoiding omissions or duplicates due to data deviations.

[0063] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for quick generation of fertilization prescription map based on multi-spectral detection, characterized in that, The method comprises the following steps: (1) Multi-source data acquisition: synchronously acquiring 450-900nm waveband crop multispectral images collected by a multi-spectral camera carried by an unmanned aerial vehicle, laboratory detected nutrient data of preset soil sampling points, real-time soil humidity data, plot boundary vector data and crop recommended fertilization model; (2) Data preprocessing: performing radiation correction, geometric correction and denoising processing on the multispectral images, calculating a vegetation index combination, and spatially registering the vegetation index data with the soil basic data and the plot boundary data to generate a raster data set in a unified coordinate system; (3) Crop nutrient requirement inversion: taking the vegetation index combination as input, constructing a crop nutrient inversion model based on a partial least squares regression (PLSR) algorithm, outputting the current nutrient content of each grid, and calculating the nutrient deficiency amount according to the crop growth period requirement; (4) Fertilization amount calculation: fusing the soil basic nutrient content and the crop nutrient deficiency amount, and calculating the unit area fertilization amount according to the following formula: Q = (D - S * N) / F Wherein: Q is the unit area fertilization amount, D is the crop nutrient deficiency amount, S is the soil effective supply amount (determined by the soil detection value and the element conversion coefficient), N is the soil nutrient utilization rate, F is the fertilizer utilization rate; (5) Prescription map generation and optimization: assigning the fertilization amount to the corresponding grid to generate an initial prescription map, eliminating the abrupt value through a spatial smoothing algorithm, and converting the prescription output into a farm operation unit prescription.

2. The method for quick generation of fertilization prescription map based on multi-spectral detection according to claim 1, characterized in that, Step (4) further comprises: Accessing real-time weather warning data, and automatically triggering dynamic correction of the fertilization amount when the predicted rainfall in the next 24 hours is > 30mm: Applying a loss prevention coefficient of 0.7 to the slope area (slope > 5°); Applying a fertilizer damage avoidance coefficient of 0.8 to the low-lying area (soil humidity > 85%); The corrected fertilization amount is updated to the prescription map generation module synchronously.

3. The method for quick generation of fertilization prescription map based on multi-spectral detection according to claim 1, characterized in that, The resolution of the multispectral images in step (1) is not higher than 0.5 meters, the soil sampling points are arranged according to a preset density, and the N, P2O5 and K2O contents of the 0-20cm plough layer soil are detected.

4. The method for quick generation of fertilization prescription map based on multi-spectral detection according to claim 1, characterized in that, The vegetation index combination includes NDVI, RENDVI and RVI, and the PLSR inversion model meets the predetermined precision requirement.

5. The method for quick generation of fertilization prescription map based on multi-spectral detection according to claim 1, characterized in that, The element conversion coefficient in step (4) is 0.6 for nitrogen, 0.5 for phosphorus and 0.7 for potassium; For the area with soil humidity > 80%, a correction coefficient of 0.8-0.9 is introduced to reduce the fertilization amount.

6. The method for quick generation of fertilization prescription map based on multi-spectral detection according to claim 1, characterized in that, In step (5), a moving window average algorithm is used for spatial smoothing processing, and the grid data is converted into an operation unit prescription according to the farm operation width, and the output format is a shp file.

7. A multispectral detection based fertilization prescription map quick generation system for implementing the method of any one of claims 1-6, characterized in that, It comprises: Data acquisition module: acquiring unmanned aerial vehicle multispectral images, soil detection data, real-time soil humidity data and plot boundary vectors; Data processing module: performing image correction, vegetation index calculation and multi-source data spatial registration; Nutrient inversion module: constructing a crop nutrient model based on a PLSR algorithm, and outputting a grid-level nutrient deficiency amount; Fertilization decision module: fusing soil basic nutrients, real-time weather warning data and terrain data to calculate a dynamically corrected fertilization amount; Prescription map generation module: generating a farm machine recognizable operation unit prescription map and outputting.

8. The multispectral detection based rapid generation of fertilizer prescription map system according to claim 7, wherein, The data acquisition module integrates a weather API interface to acquire rainfall and temperature forecast data in real time.

9. The multispectral detection based rapid generation of fertilizer prescription map system according to claim 7, wherein, The fertilization decision module has an element conversion coefficient library, a regional correction rule library and a user-defined interface.

10. The multispectral detection based quick generation of fertilization prescription map system according to claim 7, characterized in that, The prescription map generation module includes a topological verification unit for detecting the matching degree of the agricultural machine operation path and the field boundary and automatically triggering data resampling for abnormal areas.

Images