Water and fertilizer cruise decision system and method based on spectral information

Abstract

The present invention discloses a water and fertilizer cruise decision system based on spectral information, including an airborne dynamic patrol system, on which a machine vision image acquisition device and a satellite positioning device are mounted, which are used to collect spectral image feature information of crops in different growth periods and acquisition position information in real time; a moisture monitoring device, which is used to collect moisture feature information in the soil of crop planting areas in different growth periods; an analysis platform, which is connected to the airborne dynamic patrol system and the moisture monitoring device in communication, and is used to receive spectral image feature information of crops in different growth periods, acquisition position information, and moisture feature information in the soil; wherein the analysis platform includes: a data receiving module, a region segmentation module, a data preprocessing module, a data matching and analysis module, and a result output module. The present invention realizes intelligent perception targeted recognition and provides decision information for automatic water and fertilizer machines in large greenhouses or large-scale open-field planting areas.

Description

Technical Field

[0001] The present invention relates to the technical field of crop water and fertilizer cruise equipment, and more specifically, to a water and fertilizer cruise decision-making system and method based on spectral information. Background Art

[0002] Crop water, fertilizer and nutrition analysis has always been the focus of crop water, fertilizer and nutrition diagnosis and fertilization decision-making. The nutritional status of crops is a comprehensive reflection of soil nutrient supply, crop demand for nutrients and crop ability to absorb nutrients. At present, the commonly used methods for analyzing crop growth conditions include soil analysis and plant tissue analysis. For example, the patent application with application number 201910126913.0 discloses a water and fertilizer decision-making system and method based on plant life characteristics, which sets up multiple sensors: water sensor, new shoot nutrient sensor, leaf nutrient sensor, trunk nutrient sensor and root nutrient sensor. The leaf water sensor, new shoot nutrient sensor, leaf nutrient sensor, trunk nutrient sensor and root nutrient sensor collect image information during plant growth, and make decisions on the water, fertilizer and medicine required for plant growth based on the collected plant images. However, this analysis system has a complex structure, and the cost of analyzing each plant is too high for large-scale greenhouses or open-field production areas, and its applicability is poor. Therefore, there is an urgent need for a crop water and fertilizer analysis system that can achieve real-time monitoring and mobile positioning patrols in large greenhouses or large-scale open-field planting areas. Summary of the invention

[0003] An object of the present invention is to solve at least the above-mentioned problems and provide a water and fertilizer cruise decision-making system based on spectral information, which realizes real-time monitoring and mobile positioning patrol in a large z planting area through an airborne dynamic patrol system, and segments different water and fertilizer status areas and water and fertilizer level classification based on image edge recognition detection, realizes intelligent perception targeted recognition, and provides decision-making information for automated water and fertilizer machines in large greenhouses or large-scale open-field planting areas.

[0004] In order to achieve these purposes and other advantages according to the present invention, a water and fertilizer cruise decision system based on spectral information is provided, comprising:

[0005] The airborne dynamic patrol system is equipped with a machine vision image acquisition device and a satellite positioning device to collect spectral image feature information of crops at different growth stages and collect location information in real time;

[0006] A moisture monitoring device, which is used to collect moisture characteristic information in the soil of crop planting areas at different growth stages;

[0007] An analysis platform, which is in communication connection with the airborne dynamic patrol system and the moisture monitoring device, and is used to receive spectral image characteristic information of crops at different growth stages, collection location information, and moisture characteristic information in the soil of the crop planting areas at different growth stages;

[0008] Among them, the analysis platform includes: a data receiving module, a region segmentation module, a data preprocessing module, a data matching and analysis module and a result output module. The data receiving module is used to receive the information transmitted by the airborne dynamic patrol system and the moisture monitoring device. The region segmentation module is used to divide the collected ground image into different regions based on the spectral image feature information received by the data receiving module. The data preprocessing module is used to preprocess the spectral image feature information data of different regions. The data matching and analysis module is used to match and analyze the data preprocessed by the data preprocessing module. The result output module displays and outputs the results analyzed and calculated by the data matching and analysis module.

[0009] Preferably, the analysis platform further comprises: a model database storing nitrogen concentration calculation models of different crops, including: a rice leaf nitrogen concentration calculation model, a wheat leaf nitrogen concentration calculation model and a corn leaf nitrogen concentration calculation model.

[0010] Preferably, the machine vision image acquisition device is a hyperspectral spectrometer.

[0011] Preferably, the moisture monitoring device is a moisture sensor.

[0012] The present invention is further implemented by a water and fertilizer cruise decision method based on spectral information. The water and fertilizer cruise decision method based on spectral information uses the water and fertilizer cruise decision based on spectral information to make a cruise decision, and specifically includes the following steps:

[0013] Step 1: Adjust the field of view of the camera probe of the machine vision image acquisition device in the airborne dynamic patrol system to 25 degrees, maintain a height of 1m from the crop planting ground to collect spectral image feature information of crops, and obtain the collection position information after the satellite positioning device in the airborne dynamic patrol system locates the collection position, and transmits the spectral image feature information and the collection position information to the data receiving module in the analysis platform;

[0014] Step 2: The moisture monitoring device collects moisture characteristic information in the soil of the crop planting area at different growth periods, and transmits the moisture characteristic information to the data receiving module in the analysis platform;

[0015] Step three, the data receiving module transmits the received spectral image feature information and moisture feature information to the region segmentation module, the region segmentation module divides the collection site into different regions based on the spectral image feature information and moisture feature information based on image edge recognition detection, the data preprocessing module preprocesses the spectral image feature information data of different regions, the data matching and analysis module matches and analyzes the data preprocessed by the data preprocessing module and calculates the nitrogen concentration of different crop leaves, and the result output module displays and outputs the results analyzed and calculated by the data matching and analysis module.

[0016] Preferably, in step 3, the region segmentation module segments the image into different regions based on edge recognition detection, including:

[0017] S1, read the image and perform noise elimination on the image;

[0018] S2, performing edge recognition detection on the denoised image according to the similarity principle of crops of the same variety, to obtain images of different regions of the same variety;

[0019] S3. According to the spectral image characteristic information and moisture characteristic information of the same variety of crops, the neighboring regions with the same spectral image characteristic information and moisture characteristic information are merged until they are divided into the minimum number of regions.

[0020] Preferably, in step 3, the data preprocessing module preprocesses the spectral image feature information, including:

[0021] The raw data is smoothed using a moving average filter;

[0022] All smoothed data were normalized.

[0023] Preferably, in step 1, the machine vision image acquisition device acquires images in a range of 350 to 2500 nm, a sampling interval of 0.6 to 1.6 nm, and each acquisition point is measured 10 times.

[0024] The present invention has at least the following beneficial effects:

[0025] 1. Through the airborne dynamic patrol system and moisture monitoring device, real-time monitoring and mobile positioning patrol can be realized in large greenhouses or large open-field planting areas.

[0026] 2. Through the regional segmentation module in the analysis platform, the collection site is divided into different areas based on image edge recognition detection according to the spectral image feature information and moisture feature information, and the large greenhouse or large-scale open-field planting area is divided into different water and fertilizer status areas and water and fertilizer level classification, to achieve intelligent perception and targeted recognition, and provide decision-making information for the automatic water and fertilizer machine.

[0027] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a structural schematic diagram of a water and fertilizer cruise decision-making system based on spectral information, which is one of the technical solutions of the present invention;

[0029] Figure 2 This is a schematic diagram of the structure of the analysis platform in the water and fertilizer cruise decision-making system based on spectral information in one of the technical solutions of the present invention;

[0030] Figure 3 This is a flow chart of a water and fertilizer cruise decision-making method based on spectral information, which is one of the technical solutions of the present invention. DETAILED DESCRIPTION

[0031] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0032] It should be understood that the terms such as “having”, “including” and “comprising” used herein do not exclude the existence or addition of one or more other elements or combinations thereof.

[0033] like Figures 1-2 As shown, the present invention provides a water and fertilizer cruise decision-making system based on spectral information, comprising:

[0034] The airborne dynamic patrol system is equipped with a machine vision image acquisition device and a satellite positioning device to collect spectral image feature information of crops at different growth stages and collect location information in real time;

[0035] A moisture monitoring device, which is used to collect moisture characteristic information in the soil of crop planting areas at different growth stages;

[0036] An analysis platform, which is in communication connection with the airborne dynamic patrol system and the moisture monitoring device, and is used to receive spectral image characteristic information of crops at different growth stages, collection location information, and moisture characteristic information in the soil of the crop planting areas at different growth stages;

[0037] Among them, the analysis platform includes: a data receiving module, a region segmentation module, a data preprocessing module, a data matching and analysis module and a result output module. The data receiving module is used to receive the information transmitted by the airborne dynamic patrol system and the moisture monitoring device. The region segmentation module is used to divide the collected ground image into different regions based on the spectral image feature information received by the data receiving module. The data preprocessing module is used to preprocess the spectral image feature information data of different regions. The data matching and analysis module is used to match and analyze the data preprocessed by the data preprocessing module. The result output module displays and outputs the results analyzed and calculated by the data matching and analysis module.

[0038] In the above-mentioned implementation scheme, the airborne dynamic patrol system may adopt a drone equipped with a machine vision image acquisition device and a satellite positioning device. The machine vision image acquisition device is a machine vision image acquisition device with a special spectral camera, such as a hyperspectral spectrometer. The satellite positioning device may adopt a GPS system. The hyperspectral spectrometer is carried on the drone through an adjustable bracket, so that the drone carrying the hyperspectral spectrometer can perform real-time mobile monitoring and positioning patrol in a large greenhouse or a large open-field planting area. The moisture monitoring device may be a soil moisture sensor, which is buried in the soil in a large greenhouse or a large open-field planting area, and is used to collect moisture characteristic information in the soil of crop planting areas at different growth periods and transmit the collected moisture characteristic information to the analysis platform. The data receiving module in the analysis platform is used to receive the spectral image feature information and collection location information of crops in different growth periods collected by the machine vision image acquisition device, as well as the moisture feature information in the soil of the crop planting area in different growth periods; the regional segmentation module divides the collected image into different areas based on the spectral image feature information, and the image segmentation is based on image edge recognition detection segmentation. The data preprocessing module preprocesses the spectral image feature information data of different areas, and the preprocessing includes smoothing and normalization. The data matching and analysis module is used to match and analyze the data preprocessed by the data preprocessing module. The matching is based on the database in the analysis platform for model matching. After matching, the nitrogen concentration and water content of different areas are calculated according to different models. The result output module displays and outputs the results analyzed and calculated by the data matching and analysis module, and finally realizes the segmentation of a large greenhouse or a large open-field planting area into different water and fertilizer status areas, performs water and fertilizer level classification, realizes intelligent perception targeted identification, and provides decision-making information for the automatic water and fertilizer machine.

[0039] The specific decision-making process includes:

[0040] Step 1: Select clear weather for measurement, the measurement time range is 10:00-14:00, adjust the camera probe field of view of the machine vision image acquisition device in the airborne dynamic patrol system to 25 degrees, maintain a height of 1m from the crop planting ground, the acquisition range is 350-2500nm, the sampling interval is 0.6-1.6nm, and the number of measurements at each acquisition point is 10 times. Collect the spectral image feature information of crops, the satellite positioning device in the airborne dynamic patrol system locates the acquisition position and obtains the acquisition position information, and transmits the spectral image feature information and the acquisition position information to the data receiving module in the analysis platform;

[0041] Step 2: The moisture monitoring device collects moisture characteristic information in the soil of the crop planting area at different growth periods, and transmits the moisture characteristic information to the data receiving module in the analysis platform;

[0042] Step three, the data receiving module transmits the received spectral image feature information and moisture feature information to the region segmentation module, the region segmentation module divides the collection site into different regions based on the spectral image feature information and moisture feature information based on image edge recognition detection, the data preprocessing module preprocesses the spectral image feature information data of different regions, the data matching and analysis module matches and analyzes the data preprocessed by the data preprocessing module and calculates the nitrogen concentration of different crop leaves, and the result output module displays and outputs the results analyzed and calculated by the data matching and analysis module.

[0043] The analysis platform obtains spectral image feature information and collection location information of crops, moisture feature information in the soil of the crop planting area through the airborne dynamic patrol system, and the regional segmentation module divides the collection area into different areas based on image edge recognition detection according to the spectral image feature information and moisture feature information. The calculated nitrogen concentration of crop leaves and moisture feature information in the soil in different areas are further transmitted to the control system. The control system controls the water and fertilizer irrigation equipment to fertilize and water the crops in different areas according to the amount of water and fertilizer required, thereby providing water and fertilizer irrigation decision information for large-scale automated water and fertilizer irrigation equipment.

[0044] In another embodiment, the analysis platform further includes: a model database storing nitrogen concentration calculation models of different crops, including: a rice leaf nitrogen concentration calculation model, a wheat leaf nitrogen concentration calculation model, and a corn leaf nitrogen concentration calculation model.

[0045] In the above implementation scheme, the rice leaf nitrogen concentration calculation model, the wheat leaf nitrogen concentration calculation model and the corn leaf nitrogen concentration calculation model are all constructed in advance and stored in the model database for crop nitrogen calculation, thereby providing decision-making information for large-scale automatic water and fertilizer machines; wherein, the establishment of the rice leaf nitrogen concentration calculation model takes the third fully expanded leaf from the top of the rice tillering period (referred to as the top three leaves) as the research object, and measures the spectral data of each leaf in the visible to near-infrared band (350-2500nm), and smoothes and normalizes the acquired spectral data to eliminate the influence of noise and dimension, and uses the principal component analysis (PCA) method to reduce the data dimension to 22 dimensions, and selects a support vector machine model based on genetic algorithm optimization parameters (specific reference: Luo Jianjun et al., Research on classification of rice nitrogen application level based on hyperspectral and parameter optimized support vector machine, Chinese Soil and Fertilizer, 2020 (5), 250-257.). The calculation model of nitrogen concentration in wheat leaves takes wheat leaves as the research object, and constructs a SPAD hyperspectral remote sensing estimation model for wheat leaves (specific reference: Ding Shaoqiang, Research on hyperspectral remote sensing estimation model of nitrogen physiological indicators of wheat of different quality types, academic master's degree thesis, 2021). The calculation model of nitrogen concentration in corn leaves takes field corn canopy leaves as the research object, and uses Fieldspec3.0 spectrometer and Greenseeker to measure the spectral reflectance and NDVI of corn leaves under different nitrogen application rates, respectively, and extracts and optimizes the NDVI values ​​to establish an optimized NDVI inversion model for nitrogen content in corn leaves (specific reference: Zhao Mingjia et al., Inversion of nitrogen content in corn leaves based on hyperspectral optimized NDVI values, Journal of China Agricultural University, 2014, 19(6), 56-61).

[0046] In another embodiment, the machine vision image acquisition device is a hyperspectral spectrometer, wherein the hyperspectral spectrometer is an ASD Field Spec FR2500 hyperspectral spectrometer, and the hyperspectral range is 350-2500nm.

[0047] In another embodiment, the moisture monitoring device is a moisture sensor. The moisture sensor is a commonly used moisture detection device with mature existing technology and wide application.

[0048] like Figure 3 As shown, the present invention also provides a water and fertilizer cruise decision system based on spectral information, which uses Figure 1 and Figure 2 The water and fertilizer cruise decision-making based on spectral information shown in the figure makes a cruise decision, which specifically includes the following steps:

[0049] Step 1: Adjust the field of view of the camera probe of the machine vision image acquisition device in the airborne dynamic patrol system to 25 degrees, maintain a height of 1m from the crop planting ground to collect spectral image feature information of crops, and obtain the collection position information after the satellite positioning device in the airborne dynamic patrol system locates the collection position, and transmits the spectral image feature information and the collection position information to the data receiving module in the analysis platform;

[0050] Step 2: The moisture monitoring device collects moisture characteristic information in the soil of the crop planting area at different growth periods, and transmits the moisture characteristic information to the data receiving module in the analysis platform;

[0051] Step three, the data receiving module transmits the received spectral image feature information and moisture feature information to the region segmentation module, the region segmentation module divides the collection site into different regions based on the spectral image feature information and moisture feature information based on image edge recognition detection, the data preprocessing module preprocesses the spectral image feature information data of different regions, the data matching and analysis module matches and analyzes the data preprocessed by the data preprocessing module and calculates the nitrogen concentration of different crop leaves, and the result output module displays and outputs the results analyzed and calculated by the data matching and analysis module.

[0052] In the above technical scheme, in step 1, the machine vision image acquisition device is an ASDField Spec FR2500 hyperspectral spectrometer, the camera probe field of view angle is 25 degrees, clear weather is selected for acquisition, the measurement time range is 10:00-14:00, and the height of 1m from the crop planting ground is maintained. The acquisition range is 350-2500nm, the sampling interval is 0.6-1.6nm, and the number of measurements at each acquisition point is 10 times. Collect the spectral image feature information of crops; the satellite positioning device uses the GPS system to locate the acquisition position and obtain the acquisition position information, and transmits the spectral image feature information and the acquisition position information to the data receiving module in the analysis platform. In step 2, a moisture sensor is used (specifically an MP406 sensor, which is arranged in a large greenhouse or a large open-field planting area, with a layout depth of 10cm and a layout range of 1 within 0.01 square meters). In step three, the data receiving module transmits the received spectral image characteristic information and moisture characteristic information to the region segmentation module. The region segmentation module divides the collection site into different regions based on the spectral image characteristic information and moisture characteristic information based on image edge recognition detection. The image segmentation of the region segmentation module specifically includes: S1, reading the image and performing denoising on the image (denoising is performed using a median filtering algorithm); S2, performing edge recognition detection on the denoised image according to the similarity principle of crops of the same variety, and obtaining images of different regions of the same variety; S3, according to the spectral image characteristic information and moisture characteristic information of the same variety of crops, merging the neighboring regions with the same spectral image characteristic information and moisture characteristic information until they are divided into the smallest number of different regions (regions with different water and fertilizer status). , according to the spectral image feature information in different areas, the data preprocessing module preprocesses the spectral image feature information data of different areas (using a moving average filter to smooth the original data; normalizing all the data after smoothing): the moving average filter is used to smooth the original data, and the window width of the average filter is set to 5 by default. The smoothing formula is: y(1) = x(1), y(2) = [x(1) + x(2) + x(3)] / 3, y(3) = [x(1) + x(2) + x(3) + x(4) + x(5)] / 5, …y(n) = [x(n-2) + x(n-1) + x(n) + x(n+1) + x(n+2)] / 5, where x is the wavelength reflectivity of the original spectral data, y is the wavelength reflectivity of the smoothed data, and n is the wavelength; the normalization formula is: y = [(y max -y min )×(xx min )] / (x max -x min )+y min , where xmax and x min are the maximum and minimum values ​​in the original spectral data, respectively, max and min They are the maximum and minimum values ​​of the spectral data after normalization, and they are 1 and -1 respectively. Among them, the smoothing process is to remove the noise in the data, and the normalization process is to improve the model operation rate and model accuracy. The pre-processed spectral image feature information is matched and analyzed by the data matching and analysis module to calculate the nitrogen concentration of different crop leaves. If the collected planting area is rice, the pre-processed spectral image feature information is matched with the rice leaf nitrogen concentration calculation model in the model database in the analysis platform, and the rice leaf nitrogen concentration is calculated. Combined with the moisture feature information and location information collected in the soil, the large greenhouse or large-scale open-field rice planting area is divided into different areas, different water and fertilizer status areas, and water and fertilizer level classification, and then the data is provided to the subsequent fertilization and irrigation system to realize intelligent perception and targeted recognition, and provide decision-making information for the automatic rice water and fertilizer machine. If the collected planting area is wheat, the pre-processed spectral image feature information is matched with the calculation model of the nitrogen concentration of wheat leaves in the model database in the analysis platform, and the nitrogen concentration of wheat leaves is calculated, and then combined with the moisture feature information and location information collected in the soil, the large greenhouse or large-scale open-field wheat planting area is divided into different areas, different water and fertilizer status areas, and water and fertilizer level classification, and then the data is provided to the subsequent fertilization and irrigation system to achieve intelligent perception and targeted identification, and provide decision-making information for the automated wheat water and fertilizer machine. If the collected planting area is corn, the pre-processed spectral image feature information is matched with the calculation model of the nitrogen concentration of corn leaves in the model database in the analysis platform, and the nitrogen concentration of corn leaves is calculated, and then combined with the moisture feature information and location information collected in the soil, the large greenhouse or large-scale open-field corn planting area is divided into different areas, different water and fertilizer status areas, and water and fertilizer level classification, and then the data is provided to the subsequent fertilization and irrigation system to achieve intelligent perception and targeted identification, and provide decision-making information for the automated corn water and fertilizer machine.

[0053] In order to enable those skilled in the art to better understand the technical solution of the present invention, the following embodiments are provided for illustration:

[0054] <Example 1>

[0055] A water and fertilizer cruise decision system based on spectral information, comprising:

[0056] An airborne dynamic patrol system is equipped with a machine vision image acquisition device and a satellite positioning device, which are used to collect spectral image feature information of crops in different growth periods and collect location information in real time; wherein, the airborne dynamic patrol system can use a drone equipped with a machine vision image acquisition device and a satellite positioning device, the machine vision image acquisition device is a machine vision image acquisition device with a special spectral camera, such as a hyperspectral spectrometer, and the satellite positioning device can use a GPS system. The hyperspectral spectrometer is carried on the drone through an adjustable bracket, so that the drone carrying the hyperspectral spectrometer can perform real-time mobile monitoring and positioning patrol at an adjustable angle in a large greenhouse or a large open-field planting area;

[0057] A moisture monitoring device, which is used to collect moisture characteristic information in the soil of a crop planting area at different growth periods; wherein the moisture monitoring device is a soil moisture sensor, which is buried in the soil in a large greenhouse or a large open-field planting area;

[0058] An analysis platform, which is in communication connection with the airborne dynamic patrol system and the moisture monitoring device, and is used to receive spectral image characteristic information of crops at different growth stages, collection location information, and moisture characteristic information in the soil of the crop planting areas at different growth stages;

[0059] Among them, the analysis platform includes: a data receiving module, a region segmentation module, a data preprocessing module, a data matching and analysis module and a result output module. The data receiving module is used to receive information transmitted by the airborne dynamic patrol system and the moisture monitoring device. The region segmentation module is used to divide the collected ground image into different regions based on the spectral image feature information received by the data receiving module. The data preprocessing module is used to preprocess the spectral image feature information data of different regions. The data matching and analysis module is used to match and analyze the data preprocessed by the data preprocessing module. The matching and calculation are based on the model database of the analysis platform (which stores nitrogen concentration calculation models of different crops, including: rice leaf nitrogen concentration calculation model, wheat leaf nitrogen concentration calculation model and corn leaf nitrogen concentration calculation model). The result output module displays and outputs the results analyzed and calculated by the data matching and analysis module.

[0060] The water and fertilizer cruise decision method of the water and fertilizer cruise decision system based on spectral information in this embodiment specifically includes the following steps:

[0061] Step 1: Use a machine vision image acquisition device (ASDField Spec FR2500 hyperspectral spectrometer) with a camera probe field of view of 25 degrees. Select clear weather for acquisition, the measurement time range is 10:00-14:00, maintain a height of 1m from the crop planting ground, the acquisition range is 350-2500nm, the sampling interval is 0.6-1.6nm, and the number of measurements at each acquisition point is 10 times. Collect the spectral image feature information of crops; the satellite positioning device uses the GPS system to locate the acquisition position and obtain the acquisition position information, and transmit the spectral image feature information and the acquisition position information to the data receiving module in the analysis platform;

[0062] Step 2: Use a moisture sensor (specifically, an MP406 sensor, which is arranged in a large greenhouse or a large open-field planting area, with a depth of 10 cm and a range of 0.01 square meters) to collect moisture characteristic information in the soil of crop planting areas at different growth periods, and transmit the moisture characteristic information to the data receiving module in the analysis platform;

[0063] Step 3: The data receiving module transmits the received spectral image feature information and moisture feature information to the region segmentation module. The region segmentation module divides the collection area into different regions based on the spectral image feature information and moisture feature information based on image edge recognition detection. The image segmentation of the region segmentation module specifically includes: S1, reading the image and performing denoising on the image (denoising is performed using a median filter algorithm); S2, performing edge recognition detection on the denoised image according to the similarity principle of crops of the same variety to obtain images of different regions of the same variety; S3, performing spectral image feature information and moisture feature information of crops of the same variety according to the spectral image feature information and moisture feature information of crops of the same variety. , merge the neighboring areas with the same spectral image characteristic information and moisture characteristic information until they are divided into the smallest number of different areas (different water and fertilizer status areas). According to the spectral image characteristic information in different areas, the data preprocessing module preprocesses the spectral image characteristic information data of different areas (using a moving average filter to smooth the original data; normalizing all the smoothed data): Smoothing uses a moving average filter to smooth the original data. The window width of the average filter is set to 5 by default. The smoothing formula is: y(1) = x(1), y(2) = [x(1) + x(2)

[0064] +x(3)] / 3, y(3)=[x(1)+x(2)+x(3)+x(4)+x(5)] / 5, …y(n)=[x(n-2)+x(n-1)+x(n)+x(n+1)+x(n+2)] / 5, where x is the wavelength reflectivity of the original spectral data, y is the wavelength reflectivity of the smoothed data, and n is the wavelength. The normalization formula is: y=[(y max -y min )×(xx min )] / (x max -x min )+y min , where x max and x min are the maximum and minimum values ​​in the original spectral data, respectively, max and min They are the maximum and minimum values ​​of the spectral data after normalization, which are 1 and -1 respectively; the preprocessed spectral image feature information is matched and analyzed by the data matching and analysis module on the data preprocessed by the data preprocessing module to calculate the nitrogen concentration of different crop leaves, and then combined with the moisture feature information and position information in the soil collected previously, the large greenhouse or large-scale open-field planting area is divided into different areas, different water and fertilizer status areas and water and fertilizer level classification, and then the data is provided to the subsequent fertilization and irrigation system to realize intelligent perception targeted recognition and provide decision-making information for the automatic rice water and fertilizer machine.

[0065] The existing water and fertilizer decision-making systems are all carried out within a small range and within the same water and fertilizer level level area, and cannot realize the full-area real-time monitoring and mobile positioning patrol in a large greenhouse or a large-scale open-field planting area. The water and fertilizer cruise decision-making system based on spectral information of the present invention realizes the full-area real-time monitoring and mobile positioning patrol in a large greenhouse or a large-scale open-field planting area through an airborne dynamic patrol system and a moisture monitoring device. Through the regional segmentation module in the analysis platform, the collection area is divided into different areas based on image edge recognition detection according to the spectral image feature information and moisture feature information, and the large greenhouse or a large-scale open-field planting area is divided into different water and fertilizer status areas and water and fertilizer level classifications, realizing intelligent perception and targeted recognition, and providing decision information for the automatic water and fertilizer machine.

[0066] The number of devices and processing scales described here are used to simplify the description of the present invention. The application, modification and variation of the water and fertilizer cruise decision-making system and decision-making method based on spectral information of the present invention are obvious to those skilled in the art.

[0067] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A water and fertilizer cruise decision-making method based on spectral information, characterized in that: A water and fertilizer cruise decision system based on spectral information is used to make cruise decisions, wherein the water and fertilizer cruise decision system includes: an airborne dynamic patrol system equipped with a hyperspectral spectrometer and a satellite positioning device for real-time collection of spectral image feature information of crops in different growth periods and collection position information; a moisture sensor for collecting moisture feature information in soil of crop planting sites at different growth periods; an analysis platform, which is communicatively connected to the airborne dynamic patrol system and the moisture sensor, for receiving spectral image feature information of crops in different growth periods, collection position information, and moisture feature information in soil of crop planting sites at different growth periods; the analysis platform includes: a data receiving module, a region segmentation module, a data preprocessing module, a data matching and analysis module, and a result output module and a model database, wherein the data receiving module is used to receive information transmitted by the airborne dynamic patrol system and the moisture sensor, the region segmentation module is used to segment the collected ground image into different regions based on the spectral image feature information received by the data receiving module, the data preprocessing module is used to preprocess the spectral image feature information data of different regions, the data matching and analysis module is used to match and analyze the data preprocessed by the data preprocessing module, and the result output module displays and outputs the results obtained by the analysis and calculation of the data matching and analysis module; the model database stores nitrogen concentration calculation models of different crops, including: a rice leaf nitrogen concentration calculation model, a wheat leaf nitrogen concentration calculation model, and a corn leaf nitrogen concentration calculation model; The water and fertilizer cruise decision-making method based on spectral information comprises the following steps: Step 1: Adjust the field of view of the camera probe of the hyperspectral spectrometer in the airborne dynamic patrol system to 25 degrees, maintain a height of 1 m from the crop planting ground to collect spectral image feature information of crops, and obtain the collection position information after the satellite positioning device in the airborne dynamic patrol system locates the collection position, and transmits the spectral image feature information and the collection position information to the data receiving module in the analysis platform; Step 2: The moisture sensor collects moisture characteristic information in the soil of the crop planting area at different growth periods, and transmits the moisture characteristic information to the data receiving module in the analysis platform; Step 3, the data receiving module transmits the received spectral image characteristic information and moisture characteristic information to the region segmentation module, the region segmentation module divides the collection area into different regions based on the spectral image characteristic information and moisture characteristic information based on image edge recognition detection, the data preprocessing module preprocesses the spectral image characteristic information data of different regions, the data matching and analysis module matches and analyzes the data preprocessed by the data preprocessing module with the nitrogen concentration calculation model of different crops in the model database to calculate the nitrogen concentration of different crop leaves, and the result output module displays and outputs the results analyzed and calculated by the data matching and analysis module; wherein the region segmentation module divides into different regions based on image edge recognition detection specifically including: reading an image and performing denoising on the image; performing edge recognition detection on the denoised image according to the similarity principle of crops of the same variety to obtain images of different regions of the same variety; according to the spectral image characteristic information and moisture characteristic information of crops of the same variety, merging neighboring regions with the same spectral image characteristic information and moisture characteristic information until they are divided into the least number of regions.

2. The water and fertilizer cruise decision-making method based on spectral information according to claim 1, characterized in that: In step 3, the data preprocessing module preprocesses the spectral image feature information including: The raw data is smoothed using a moving average filter; All smoothed data were normalized.

3. The water and fertilizer cruise decision-making method based on spectral information according to claim 1, characterized in that: In step 1, the acquisition range of the hyperspectral spectrometer is 350-2500 nm, the sampling interval is 0.6-1.6 nm, and the number of measurements at each acquisition point is 10 times.

Images