Dry-farming oat growth vigor monitoring method and system based on hyperspectral remote sensing of unmanned aerial vehicle

Abstract

The invention provides a dry-farming oat growth vigor monitoring method and system based on unmanned aerial vehicle hyperspectral remote sensing, and relates to the technical field of dry-farming oat growth vigor monitoring. The reflectivity of each image pixel under different spectral band conditions is calculated through the absolute radiation brightness and the known reflectivity of the reference plate, the average reflectivity of the unit grids in different spectral bands is obtained, a linear product is adopted to construct a growth score index, the growth score index of each grid unit is clustered by adopting a K-means clustering method, and the average reflectivity of each grid unit is calculated. And outputting the clustering condition of the growth vigor of the dry farming oats of the grid units, and setting a judgment threshold according to the clustering condition of the growth vigor of the dry farming oats to realize growth vigor classification. A reference plate is introduced to calibrate the reflectivity, the pixel-level absolute reflectivity is calculated through a radiation transfer equation, and the influence of illumination condition changes is eliminated; k-means clustering is adopted to automatically divide growth vigor levels, and a threshold value can be generated in a self-adaptive mode based on data distribution.

Description

Technical Field

[0001] This invention relates to the field of dryland oat growth monitoring and processing technology, specifically to a method and system for monitoring dryland oat growth based on UAV hyperspectral remote sensing. Background Technology

[0002] Traditional methods for monitoring the growth of dryland oats typically rely on manual surveys or low-resolution remote sensing imagery. These methods have limitations in both time efficiency and spatial resolution, failing to reflect crop growth status in a timely and accurate manner, thus hindering farmers from making timely agricultural management decisions. With the rapid development of drone technology, hyperspectral remote sensing technology offers a new solution for crop monitoring. By acquiring high-resolution spectral images, it is possible to more accurately assess crop growth status and its environmental stress response.

[0003] In the prior art, CN116824400A discloses a method for using drones to photograph crops in different areas. This method obtains remote sensing images of different sub-regions, stitches them together to form a comprehensive remote sensing image of the crop, and analyzes this image to obtain information on the crop's growth distribution and thus determine the crop's growth requirements in different sub-regions. However, this method does not consider the impact of bare soil on vegetation, nor does it consider the emissivity of vegetation under different spectral conditions. Consequently, it does not use multiple index indicators to reflect the crop's growth trend, nor does it set appropriate thresholds to classify the crop's growth status. Therefore, there is an urgent need for a method for monitoring the growth of dryland oats based on UAV remote sensing.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing includes the following steps: S1: The boundary of the target monitoring dryland oat planting area is marked. The planting area is divided into several identical rectangular grid units according to the marked boundary. The UAV is used to acquire the spectral image of each grid unit and preprocess the spectral image. S2: Filter the reflectance bands of the preprocessed spectral image. For each grid cell, obtain the average reflectance of all pixels in different bands to obtain the average reflectance data of the grid cell in different bands. Classify all grid cells according to the set band threshold and mark the grid cells as bare soil areas or vegetation-covered areas. S3: Based on the spectral images of all grid cells marked as bare soil areas, dynamically calculate the adaptive parameters of the soil background. For grid cells marked as vegetated areas, construct vegetation spectral indices and water stress indices using multispectral coupling as indicators for monitoring the growth of dryland oats. Correct the vegetation spectral indices and water stress indices corresponding to the grid cells in the vegetated areas using the adaptive parameters of the soil background. S4: Based on the vegetation spectral index and water stress index corrected by the grid cells of each vegetation cover area, a growth score index is constructed by linear multiplication, and an adaptive parameter-corrected growth score index for soil background is generated. S5: K-means clustering is used to cluster the growth score index of each grid cell in the vegetation cover area, and the clustering of the growth of dryland oats in the grid cells of the vegetation cover area is output. Based on the clustering of the growth of dryland oats, a judgment threshold is set to achieve growth classification.

[0007] Furthermore, the spectral image includes reflectance spectral information from multiple pixels. This spectral information includes the correspondence between wavelength and reflectance values. The method for preprocessing the spectral image is as follows: For each pixel in the spectral image of each grid cell, the reflectance spectral curve of that pixel is extracted. Based on the set window length, for each band in the spectral curve of that pixel, the reflectance values ​​of several bands before and after it are taken. The average reflectance within the window is used as the smoothed output value of the current band. The original reflectance value is replaced with the smoothed output value to complete the preprocessing of the reflectance spectral curve of each pixel.

[0008] Furthermore, the selected reflectivity bands are as follows: , , and For each preprocessed spectral image, for the reflectance band of the selected wave, all pixels in the spectral image are traversed, the reflectance values ​​of all pixels in each selected wave are collected, and the average of all collected pixel reflectance values ​​is calculated to obtain the average reflectance of the spectral image in that wave band.

[0009] For the spectral image corresponding to each grid cell, the average reflectance of the four target bands is calculated sequentially to obtain the average reflectance of each grid cell in the four target bands. , , and Average reflectance across four bands; The average reflectance of each grid cell across the four bands is converted to absolute radiance using the following formula: ; in, Indicates the first Each grid cell in Mean reflectance in the band, and They represent Gain coefficient and bias term for band reflectivity, Indicates the first Each grid cell in Absolute radiance in the band, This indicates the wavelength value of the selected band; The average reflectance data of each grid cell in different bands is obtained by combining the absolute radiance of the reference plate with the average reflectance of each grid cell across four bands, according to the formula: ; in, No. Each grid cell in Average reflectance data in the band, and They represent The radiance and calibrated reflectance of the reference plate in the specified band, and the average reflectance data for each grid cell are included in... , , and Average reflectance data for four bands.

[0010] Furthermore, near-infrared sensors were installed on the vegetation. The lower limit threshold of spectral reflectance and red light The upper limit threshold of spectral reflectance is used to classify the entire planting area. Images with reflectance within the threshold range are identified as bare soil areas, while images with reflectance outside the threshold range are identified as vegetation-covered areas.

[0011] For each grid cell, obtain its position in... bands and Average reflectance data under each band, respectively set Lower limit threshold of reflectivity for the band and The upper limit threshold for reflectivity in a band is determined according to the following classification rules: When the grid cell The average reflectivity of the band is higher than or equal to the set lower limit threshold for reflectivity, and When the average reflectance of a band is lower than or equal to the set upper limit threshold for reflectance, the grid cell is marked as a vegetation-covered area; otherwise, the grid cell is marked as a bare soil area. By performing the above-mentioned judgment on all grid cells in sequence, the classification of bare soil areas and vegetation-covered areas can be achieved.

[0012] Furthermore, based on the spectral images of all grid cells marked as bare soil areas, adaptive parameters of the soil background are dynamically calculated. The specific steps are as follows: For all grid cells marked as bare soil areas, collect the spectral image corresponding to each grid cell, including... , , and Average reflectance data for four spectral bands; For all grid cells in the exposed soil area, the average reflectance values ​​under each band are statistically analyzed to form a set of spectral characteristics of the exposed soil area; Based on the spectral feature set of exposed soil areas, adaptive parameters of the soil background are calculated using statistical analysis methods. These adaptive parameters include the mean and standard deviation of the average soil reflectance in each band. The specific steps are as follows: For each band, the mean of the average reflectance of all grid cells in the exposed soil area is calculated and used as the adaptive mean parameter of the soil background in that band. For each target band, the standard deviation of the average reflectance of all grid cells in the exposed soil area is calculated and used as the adaptive dispersion parameter of the soil background in that band.

[0013] Furthermore, based on the average reflectance of the grid cells, multispectral coupling was used to construct vegetation spectral indices and water stress indices for the grid cells marked as vegetation-covered areas, serving as indicators for monitoring the growth of dryland oats. Specifically: ; ; in, Indicates the first Vegetation spectral index of each grid cell; Indicates the first Moisture stress index of each grid cell; This represents the number of grid indices for the entire planting area; Indicates band The next Average reflectance of each grid cell; Indicates band The next Average reflectance of each grid cell; Indicates band The next Average reflectance of each grid cell.

[0014] Furthermore, the vegetation spectral index and water stress index corresponding to the grid of the vegetation cover area are corrected by adaptive parameters of the soil background. The specific steps are as follows: For grid units in vegetated areas For bands reflectivity The corrected value is: ; in, Indicates band The corrected average reflectance; Indicates band The mean soil reflectance; Indicates band The standard deviation of soil reflectance; Based on the corrected average reflectance of each band, adaptive parameter-corrected vegetation spectral index and water stress index for soil background are generated. The specific steps are as follows: ; ; in, Indicates the first Corrected vegetation spectral index for each grid cell; Indicates band Average reflectance after downcorrection; Indicates band Average reflectance after downcorrection; Indicates the first The corrected water stress index for each grid cell; For the For each grid cell, a growth score index is constructed based on the corrected vegetation spectral index and water stress index: ; Indicates the first The growth score index of each grid cell; The score indicates the inhibitory effect of water stress on growth, determined based on expert scoring.

[0015] Furthermore, K-means clustering was used to cluster the growth score index of each grid point, outputting the growth classification of dryland oats in the entire wheat field, specifically: Three initial cluster center values ​​are randomly selected as the starting centers for each category. For the growth score index of each grid cell in the vegetation cover area, the distance between it and each cluster center is calculated, and it is assigned to the category of the nearest cluster center. For each category, calculate the mean of the growth score index of all grid cells assigned to that category, and use it as the new cluster center; Based on the updated cluster centers, the distance between each grid cell in the vegetation cover area and the updated cluster centers is calculated repeatedly according to the growth score index of each grid cell. The cells are then assigned to the category of the nearest cluster center until the cluster centers no longer change, thus completing the clustering process.

[0016] Furthermore, based on the clustering of growth patterns of dryland oats, a judgment threshold is set to achieve growth classification, specifically as follows: The final growth score index is divided into: There are clusters, and the cluster center for each cluster is . And arranged in ascending order: ; in, , These represent the smallest, medium, and largest cluster centers of the growth score index, respectively. Calculate the dividing point of growth separately: ; ; in, This indicates the dividing line between weak and normal growth. This indicates the dividing line between normal and vigorous growth. when Less than or equal to At that time, it was determined that the dryland oats were in a weak growth condition; when Greater than and less than or equal to At that time, it was determined that the dryland oats were growing normally; when Greater than At that time, it was determined that the dryland oats were growing vigorously.

[0017] The present invention also provides a system for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing, the monitoring system being used to perform the above-mentioned monitoring method, including: The data acquisition module is used to mark the boundaries of the target monitoring dryland oat planting area. Based on the marked boundaries, the planting area is divided into several identical rectangular grid units. The UAV is used to acquire the spectral images of each grid unit and preprocess the spectral images. The grid processing module is used to filter the reflectance bands of the preprocessed spectral image. For each grid cell, the average reflectance of all pixels in different bands is obtained to obtain the average reflectance data of the grid cell in different bands. All grid cells are classified according to the set band threshold and marked as bare soil area or vegetation cover area. The index construction module is used to dynamically calculate the adaptive parameters of the soil background based on the spectral images of all grid cells marked as bare soil areas. For grid cells marked as vegetated areas, multispectral coupling is used to construct vegetation spectral indices and water stress indices as indicators for monitoring the growth of dryland oats. The adaptive parameters of the soil background are used to correct the vegetation spectral indices and water stress indices corresponding to the grid cells in the vegetated areas. The growth scoring module is used to construct a growth scoring index based on the vegetation spectral index and water stress index corrected by the grid cells of each vegetation cover area, and to generate a growth scoring index after adaptive parameter correction of the soil background. The growth classification module is used to cluster the growth score index of each grid unit in the vegetation coverage area using the K-means clustering method, outputting the clustering status of dryland oats in the grid units of the vegetation coverage area, and setting a judgment threshold based on the clustering status of dryland oats to achieve growth classification.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention addresses the shortcomings of traditional static thresholds in adapting to complex farmland environments by dynamically compensating for soil background interference in vegetation spectral indices using adaptive parameters based on the brightness index of bare soil areas. It constructs a growth vigor scoring index through the linear product of vegetation spectral indices and water stress indices, simultaneously characterizing vegetation cover and water stress levels, overcoming the limitations of single-index analysis. Furthermore, it employs K-means clustering combined with adaptive scoring thresholds to achieve locally adaptive classification of growth levels, improving classification accuracy compared to global threshold classification methods, particularly enhancing robustness in sparsely vegetated areas. Finally, it couples the corrected vegetation indices with the water stress index through linear product to construct a composite growth vigor scoring index, overcoming the limitations of traditional methods that solely rely on vegetation cover. This index simultaneously reflects crop biomass and drought resistance, providing multidimensional evidence for accurate growth assessment. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a graph showing the relationship between the growth score index and the corrected vegetation spectral index. Figure 3 A graph showing the relationship between the growth score index and the corrected water stress index; Figure 4 This is a block diagram of the overall system structure of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0021] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0022] Example: Please see Figures 1-3 The present invention provides a technical solution: A method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing includes the following steps: S1: The boundary of the target monitoring dryland oat planting area is marked. The planting area is divided into several identical rectangular grid units according to the marked boundary. The UAV is used to acquire the spectral image of each grid unit and preprocess the spectral image. When marking the boundaries of the target dryland oat planting area, high-resolution remote sensing images of the area can first be acquired using high-precision GPS equipment and drones. Then, the images are analyzed using remote sensing image processing software, and the outline of the oat planting area is identified and the boundaries are marked using manual or automated edge detection algorithms. After the marking is completed, the planting area is divided into multiple rectangular grid units of the same size using GIS software to ensure that each grid unit can uniformly cover the entire planting area and facilitate the subsequent acquisition of spectral image data by drones within each grid. This division will provide a standardized spatial framework for subsequent spectral analysis and crop growth monitoring.

[0023] The spectral image includes reflectance spectral information of multiple pixels. The spectral information includes the correspondence between wavelength and reflectance values. The method for preprocessing the spectral image is as follows: For each pixel in the spectral image of each grid cell, the reflectance spectral curve of that pixel is extracted. Based on the set window length, for each band in the spectral curve of that pixel, the reflectance values ​​of several bands before and after it are taken. The average reflectance within the window is used as the smoothed output value of the current band. The original reflectance value is replaced with the smoothed output value to complete the preprocessing of the reflectance spectral curve of each pixel.

[0024] Spectral data acquisition is often affected by environmental factors and sensor noise, leading to fluctuations in reflectance values. Smoothing the reflectance values ​​for each band by setting a window can effectively reduce this random noise, thereby improving the signal-to-noise ratio. Replacing the original reflectance values ​​with smoothed output values ​​helps eliminate inconsistencies caused by small-range fluctuations across different wavelengths, resulting in smoother and more consistent spectral curves. Using the average value within the window can better highlight important features or trends in the spectral curve, such as vegetation indices or soil properties. This smoothing process helps analysts more easily identify and extract features related to important information such as crop health and soil characteristics. In subsequent classification and analysis, smoothed spectral data provides more accurate information, helping to improve the identification accuracy of different categories, such as vegetated areas and bare soil areas.

[0025] S2: Filter the reflectance bands of the preprocessed spectral image. For each grid cell, obtain the average reflectance of all pixels in different bands to obtain the average reflectance data of the grid cell in different bands. Classify all grid cells according to the set band threshold and mark the grid cells as bare soil areas or vegetation-covered areas. The selected reflectance bands are respectively , , and For each preprocessed spectral image, for the reflectance band of the selected wave, all pixels in the spectral image are traversed, the reflectance values ​​of all pixels in each selected wave are collected, and the average of all collected pixel reflectance values ​​is calculated to obtain the average reflectance of the spectral image in that wave band.

[0026] Filtering reflectivity bands , , and The reason is that these bands can effectively reflect the spectral characteristics of vegetation and soil, and have a strong ability to distinguish them. The wavelength is in the visible light range and is mainly used to obtain the green reflectance characteristics of vegetation, which helps to assess the health status of vegetation. In the near-infrared region, the wavelength can effectively distinguish the growth of vegetation, and the reflectance of vegetation shows obvious changes in this wavelength. The bands are used to identify moisture content and soil properties, providing information on soil moisture and vegetation water stress; while The bands are used to acquire blue light reflectance characteristics, which can be used to analyze the photosynthetic activity of plants. Combining the reflectance data of these bands helps to accurately classify grid units as bare soil areas or vegetation-covered areas, thereby supporting subsequent agricultural monitoring and management decisions.

[0027] For the spectral image corresponding to each grid cell, the average reflectance of the four target bands is calculated sequentially to obtain the average reflectance of each grid cell in the four target bands. , , and Average reflectance across four bands; The average reflectance of each grid cell across the four bands is converted to absolute radiance using the following formula: ; in, Indicates the first Each grid cell in Mean reflectance in the band, and They represent Gain coefficient and bias term for band reflectivity, Indicates the first Each grid cell in Absolute radiance in the band, This indicates the wavelength value of the selected band; Converting the average reflectance of each grid cell across the four bands to absolute radiance aims to ensure data consistency and comparability under different conditions. Absolute radiance can more accurately characterize the true radiation properties of an object, facilitating subsequent analysis and comparison. (Formula) middle, Indicates the first Each grid cell in Average reflectance under the specified band; express The gain coefficient of band reflectivity is used as a linear scaling factor to convert reflectivity values ​​into radiance. express The bias term for band reflectance corrects for background noise and other ineffective signals in the system. Using the above formula, reflectance under the same band can be consistently converted to absolute radiance, better supporting subsequent remote sensing analysis and applications. The average reflectance data of each grid cell in different bands is obtained by combining the absolute radiance of the reference plate with the average reflectance of each grid cell across four bands, according to the formula: ; in, No. Each grid cell in Average reflectance data in the band, and They represent The radiance and calibrated reflectance of the reference plate in the specified band, and the average reflectance data for each grid cell are included in... , , and Average reflectance data for four bands.

[0028] The purpose of the above formula is to perform radiometric calibration using the absolute radiance of the reference plate, thereby improving the accuracy and consistency of reflectance data for each grid cell across different spectral bands. By ratioing the radiance of the grid cell to the reference value, measurement errors caused by environmental changes, sensor characteristics, and other factors can be corrected, thus standardizing the reflectance data of each grid cell into a relatively consistent form. This ensures comparability between different spectral bands and different geographical regions, thereby enhancing the reliability of subsequent analysis and classification.

[0029] Near-infrared settings for vegetation The lower limit threshold of spectral reflectance and red light The upper limit threshold of spectral reflectance is used to classify the entire planting area. Images with reflectance within the threshold range are identified as bare soil areas, while images with reflectance outside the threshold range are identified as vegetation-covered areas.

[0030] For each grid cell, obtain its position in... bands and Average reflectance data under each band, respectively set Lower limit threshold of reflectivity for the band and The upper limit threshold for reflectivity in a band is determined according to the following classification rules: When the grid cell The average reflectivity of the band is higher than or equal to the set lower limit threshold for reflectivity, and When the average reflectance of a band is lower than or equal to the set upper limit threshold for reflectance, the grid cell is marked as a vegetation-covered area; otherwise, the grid cell is marked as a bare soil area. By performing the above-mentioned judgment on all grid cells in sequence, the classification of bare soil areas and vegetation-covered areas can be achieved.

[0031] In the above process, due to the influence of leaf structure and biochemical composition, vegetation cover areas exhibit near-infrared wavelengths. The red band has high reflectivity, while the red band is... This results in lower reflectivity, while exposed soil areas exhibit lower reflectivity in the near-infrared band and relatively higher reflectivity in the red band.

[0032] Therefore, by setting a lower threshold for reflectance in the near-infrared band and an upper threshold for reflectance in the red band, it is possible to effectively distinguish between vegetated areas and bare soil areas: when the near-infrared reflectance of a grid cell is higher than the threshold and the red reflectance is lower than the threshold, it indicates that the area is mainly covered by vegetation; otherwise, it indicates that it is a bare soil area.

[0033] S3: Based on the spectral images of all grid cells marked as bare soil areas, dynamically calculate the adaptive parameters of the soil background. For grid cells marked as vegetated areas, construct vegetation spectral indices and water stress indices using multispectral coupling as indicators for monitoring the growth of dryland oats. Correct the vegetation spectral indices and water stress indices corresponding to the grid cells in the vegetated areas using the adaptive parameters of the soil background. Based on the spectral images of all grid cells marked as bare soil areas, adaptive parameters of the soil background are dynamically calculated. The specific steps are as follows: For all grid cells marked as bare soil areas, collect the spectral image corresponding to each grid cell, including... , , and Average reflectance data for four spectral bands; For all grid cells in the exposed soil area, the average reflectance values ​​under each band are statistically analyzed to form a set of spectral characteristics of the exposed soil area; Based on the spectral feature set of exposed soil areas, adaptive parameters of the soil background are calculated using statistical analysis methods. These adaptive parameters include the mean and standard deviation of the average soil reflectance in each band. The specific steps are as follows: For each band, the mean of the average reflectance of all grid cells in the exposed soil area is calculated and used as the adaptive mean parameter of the soil background in that band. For each target band, the standard deviation of the average reflectance of all grid cells in the exposed soil area is calculated and used as the adaptive dispersion parameter of the soil background in that band.

[0034] Based on the average reflectance of grid cells, multispectral coupling was used to construct vegetation spectral indices and water stress indices for grid cells marked as vegetation-covered areas, serving as indicators for monitoring the growth of dryland oats. Specifically: ; ; in, Indicates the first Vegetation spectral index of each grid cell; Indicates the first Moisture stress index of each grid cell; This represents the number of grid indices for the entire planting area; Indicates band The next Average reflectance of each grid cell; Indicates band The next Average reflectance of each grid cell; Indicates band The next Average reflectance of each grid cell.

[0035] The vegetation spectral index and water stress index corresponding to the grid of the vegetation cover area are corrected by adaptive parameters of soil background. The specific steps are as follows: For grid units in vegetated areas For bands reflectivity The corrected value is: ; in, Indicates band The corrected average reflectance; Indicates band The mean soil reflectance; Indicates band The standard deviation of soil reflectance; Based on the corrected average reflectance of each band, adaptive parameter-corrected vegetation spectral index and water stress index for soil background are generated. The specific steps are as follows: ; ; in, Indicates the first Corrected vegetation spectral index for each grid cell; Indicates band Average reflectance after downcorrection; Indicates band Average reflectance after downcorrection; Indicates the first The corrected water stress index for each grid cell; In the above formula, since The band falls within the near-infrared spectral range, where healthy vegetation tissue exhibits high reflectivity. The band corresponds to the red light spectrum, which is strongly absorbed by vegetation chlorophyll. Among them, high near-infrared reflectivity reflects the integrity of vegetation cell structure and biomass, while strong infrared absorption can characterize chlorophyll content and photosynthetic efficiency. By subtracting the mean soil reflectivity and standardizing, reflection interference from bare soil or sparse vegetation areas is eliminated, enhancing the purity of vegetation signals.

[0036] The wavelength is located in the shortwave infrared range and is highly sensitive to vegetation moisture content. The near-infrared band is less affected by water and reflects the stability of vegetation structure; the ratio of the two can quantify the degree of water stress. When water is deficient, the cellular structure of vegetation is damaged, and the reflectivity of the short-wave infrared spectrum increases significantly. The near-infrared band serves as a stable near-infrared reflectance reference, eliminating interference from changes in vegetation cover. Standardization eliminates the background influence of soil moisture differences on the reflectance of the two bands, making the index more accurately reflect the internal moisture status of vegetation.

[0037] S4: Based on the vegetation spectral index and water stress index corrected by the grid unit of each vegetation cover area, a growth score index is constructed by linear product, and an adaptive parameter-corrected growth score index for soil background is generated. For the For each grid cell, a growth score index is constructed based on the corrected vegetation spectral index and water stress index: ; Indicates the first The growth score index of each grid cell; The score indicates the inhibitory effect of water stress on growth, determined based on expert scoring.

[0038] Dependent variable Indicates the first The growth score index of each grid cell reflects the overall health of vegetation under growth status and water stress conditions, and is obtained through a corrected vegetation spectral index. Water stress index The ratio is obtained by normalization; through Adjust the inhibitory weight of water stress on growth to adapt to the drought resistance characteristics of different regions or crops; Characterizing vegetation cover density and photosynthetic activity, and Positive correlation; the higher the vegetation cover, the better. The larger the growth score index The closer the value is to 1, the more positive the contribution of increased vegetation cover to the growth score. Located in the denominator term, and with Multiplication increases the denominator, indicating more severe water stress. The larger the denominator, the larger the value. right The stronger the inhibition; The larger the value, the more significant the negative impact of water stress on the growth score.

[0039] In the above embodiments, the change data of 10 groups of growth score indices, the corrected vegetation spectral index, and the water stress index were selected respectively, as shown in Tables 1 and 2: Table 1: Relationship between growth score index and corrected vegetation spectral index

[0040] As shown in Table 1, the corrected water stress index remains unchanged and In this case, as the corrected vegetation spectral index increases, the growth score index also increases.

[0041] Table 2: Relationship between Growth Score Index and Corrected Water Stress Index

[0042] As shown in Table 2, the corrected vegetation spectral index remains unchanged and In this case, as the corrected water stress index increases, the growth score index decreases accordingly.

[0043] S5: K-means clustering is used to cluster the growth score index of each grid cell in the vegetation cover area, and the clustering of the growth of dryland oats in the grid cells of the vegetation cover area is output. Based on the clustering of the growth of dryland oats, a judgment threshold is set to achieve growth classification.

[0044] The K-means clustering method is used to cluster the growth score index of each grid point, and the growth classification of dryland oats in the entire wheat field is output. Specifically: Three initial cluster center values ​​are randomly selected as the starting centers for each category. For the growth score index of each grid cell in the vegetation cover area, the distance between it and each cluster center is calculated, and it is assigned to the category of the nearest cluster center. For each category, calculate the mean of the growth score index of all grid cells assigned to that category, and use it as the new cluster center; Based on the updated cluster centers, the distance between each grid cell in the vegetation cover area and the updated cluster centers is calculated repeatedly according to the growth score index of each grid cell. The cells are then assigned to the category of the nearest cluster center until the cluster centers no longer change, thus completing the clustering process.

[0045] The method of setting a judgment threshold based on the clustering of dryland oat growth to achieve growth classification is as follows: The final growth score index is divided into: There are clusters, and the cluster center for each cluster is . And arranged in ascending order: ; in, , These represent the smallest, medium, and largest cluster centers of the growth score index, respectively. Calculate the dividing point of growth separately: ; ; in, This indicates the dividing line between weak and normal growth. This indicates the dividing line between normal and vigorous growth. when Less than or equal to At that time, it was determined that the dryland oats were in a weak growth condition; when Greater than and less than or equal to At that time, it was determined that the dryland oats were growing normally; when Greater than At that time, it was determined that the dryland oats were growing vigorously.

[0046] Please see Figure 4 The present invention also provides a monitoring system for the growth of dryland oats based on UAV hyperspectral remote sensing, the monitoring system being used to perform the above-mentioned monitoring method, including: The data acquisition module is used to mark the boundaries of the target monitoring dryland oat planting area. Based on the marked boundaries, the planting area is divided into several identical rectangular grid units. The UAV is used to acquire the spectral images of each grid unit and preprocess the spectral images. The grid processing module is used to filter the reflectance bands of the preprocessed spectral image. For each grid cell, the average reflectance of all pixels in different bands is obtained to obtain the average reflectance data of the grid cell in different bands. All grid cells are classified according to the set band threshold and marked as bare soil area or vegetation cover area. The index construction module is used to dynamically calculate the adaptive parameters of the soil background based on the spectral images of all grid cells marked as bare soil areas. For grid cells marked as vegetated areas, multispectral coupling is used to construct vegetation spectral indices and water stress indices as indicators for monitoring the growth of dryland oats. The adaptive parameters of the soil background are used to correct the vegetation spectral indices and water stress indices corresponding to the grid cells in the vegetated areas. The growth scoring module is used to construct a growth scoring index based on the vegetation spectral index and water stress index corrected by the grid cells of each vegetation cover area, and to generate a growth scoring index after adaptive parameter correction of the soil background. The growth classification module is used to cluster the growth score index of each grid unit in the vegetation coverage area using the K-means clustering method, outputting the clustering status of dryland oats in the grid units of the vegetation coverage area, and setting a judgment threshold based on the clustering status of dryland oats to achieve growth classification.

[0047] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0048] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0049] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0050] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing, characterized by the following steps: include: S1: The boundary of the target monitoring dryland oat planting area is marked. The planting area is divided into several identical rectangular grid units according to the marked boundary. The UAV is used to acquire the spectral image of each grid unit and preprocess the spectral image. S2: Filter the reflectance bands of the preprocessed spectral image. For each grid cell, obtain the average reflectance of all pixels in different bands to obtain the average reflectance data of the grid cell in different bands. Classify all grid cells according to the set band threshold and mark the grid cells as bare soil areas or vegetation-covered areas. S3: Based on the spectral images of all grid cells marked as bare soil areas, dynamically calculate the adaptive parameters of the soil background. For grid cells marked as vegetated areas, construct vegetation spectral indices and water stress indices using multispectral coupling as indicators for monitoring the growth of dryland oats. Correct the vegetation spectral indices and water stress indices corresponding to the grid cells in the vegetated areas using the adaptive parameters of the soil background. S4: Based on the vegetation spectral index and water stress index corrected by the grid cells of each vegetation cover area, a growth score index is constructed by linear multiplication, and an adaptive parameter-corrected growth score index for soil background is generated. S5: K-means clustering is used to cluster the growth score index of each grid cell in the vegetation cover area, and the clustering of the growth of dryland oats in the grid cells of the vegetation cover area is output. Based on the clustering of the growth of dryland oats, a judgment threshold is set to achieve growth classification.

2. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 1, characterized in that, The spectral image includes reflectance spectral information of multiple pixels. The spectral information includes the correspondence between wavelength and reflectance values. The method for preprocessing the spectral image is as follows: For each pixel in the spectral image of each grid cell, the reflectance spectral curve of that pixel is extracted. Based on the set window length, for each band in the spectral curve of that pixel, the reflectance values ​​of several bands before and after it are taken. The average reflectance within the window is used as the smoothed output value of the current band. The original reflectance value is replaced with the smoothed output value to complete the preprocessing of the reflectance spectral curve of each pixel.

3. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 1, characterized in that, The selected reflectance bands are respectively , , and For each preprocessed spectral image, for the reflectance band of the selected wave, all pixels in the spectral image are traversed, the reflectance values ​​of all pixels in each selected wave are collected, and the average of all collected pixel reflectance values ​​is calculated to obtain the average reflectance of the spectral image in that wave band. For the spectral image corresponding to each grid cell, the average reflectance of the four target bands is calculated sequentially to obtain the average reflectance of each grid cell in the four target bands. , , and Average reflectance across four bands; The average reflectance of each grid cell across the four bands is converted to absolute radiance using the following formula: ； in, Indicates the first Each grid cell in Mean reflectance in the band, and They represent Gain coefficient and bias term for band reflectivity, Indicates the first Each grid cell in Absolute radiance in the band, This indicates the wavelength value of the selected band; The average reflectance data of each grid cell in different bands is obtained by combining the absolute radiance of the reference plate with the average reflectance of each grid cell across four bands, according to the formula: ； in, No. Each grid cell in Average reflectance data in the band, and They represent The radiance and calibrated reflectance of the reference plate in the specified band, and the average reflectance data for each grid cell are included in... , , and Average reflectance data for four bands.

4. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 1, characterized in that, Near-infrared settings for vegetation The lower limit threshold of spectral reflectance and red light The upper limit threshold of spectral reflectance is used to classify the entire planting area based on the set threshold. Images with reflectance within the threshold range are identified as bare soil areas, while images with reflectance outside the threshold range are identified as vegetation-covered areas. For each grid cell, obtain its position in... bands and Average reflectance data under each band, respectively set Lower limit threshold of reflectivity for the band and The upper limit threshold for reflectivity in a band is determined according to the following classification rules: When the grid cell The average reflectivity of the band is higher than or equal to the set lower limit threshold for reflectivity, and When the average reflectance of a band is lower than or equal to the set upper limit threshold for reflectance, the grid cell is marked as a vegetation-covered area; otherwise, the grid cell is marked as a bare soil area. By performing the above-mentioned judgment on all grid cells in sequence, the classification of bare soil areas and vegetation-covered areas can be achieved.

5. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 3, characterized in that, Based on the spectral images of all grid cells marked as bare soil areas, adaptive parameters of the soil background are dynamically calculated. The specific steps are as follows: For all grid cells marked as bare soil areas, collect the spectral image corresponding to each grid cell, including... , , and Average reflectance data for four spectral bands; For all grid cells in the exposed soil area, the average reflectance values ​​under each band are statistically analyzed to form a set of spectral characteristics of the exposed soil area; Based on the spectral feature set of exposed soil areas, adaptive parameters of the soil background are calculated using statistical analysis methods. These adaptive parameters include the mean and standard deviation of the average soil reflectance in each band. The specific steps are as follows: For each band, the mean of the average reflectance of all grid cells in the exposed soil area is calculated and used as the adaptive mean parameter of the soil background in that band. For each target band, the standard deviation of the average reflectance of all grid cells in the exposed soil area is calculated and used as the adaptive dispersion parameter of the soil background in that band.

6. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 3, characterized in that, Based on the average reflectance of grid cells, multispectral coupling was used to construct vegetation spectral indices and water stress indices for grid cells marked as vegetation-covered areas, serving as indicators for monitoring the growth of dryland oats. Specifically: ； ； in, Indicates the first Vegetation spectral index of each grid cell; Indicates the first Moisture stress index of each grid cell; This represents the number of grid indices for the entire planting area; Indicates band The next Average reflectance of each grid cell; Indicates band The next Average reflectance of each grid cell; Indicates band The next Average reflectance of each grid cell.

7. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 6, characterized in that, The vegetation spectral index and water stress index corresponding to the grid of the vegetation cover area are corrected by adaptive parameters of soil background. The specific steps are as follows: For grid units in vegetated areas For bands reflectivity The corrected value is: ； in, Indicates band The corrected average reflectance; Indicates band The mean soil reflectance; Indicates band The standard deviation of soil reflectance; Based on the corrected average reflectance of each band, adaptive parameter-corrected vegetation spectral index and water stress index for soil background are generated. The specific steps are as follows: ； ； in, Indicates the first Corrected vegetation spectral index for each grid cell; Indicates band Average reflectance after downcorrection; Indicates band Average reflectance after downcorrection; Indicates the first The corrected water stress index for each grid cell; For the For each grid cell, a growth score index is constructed based on the corrected vegetation spectral index and water stress index: ； Indicates the first The growth score index of each grid cell; The score indicates the inhibitory effect of water stress on growth, determined based on expert scoring.

8. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 1, characterized in that, The K-means clustering method is used to cluster the growth score index of each grid point, and the growth classification of dryland oats in the entire wheat field is output. Specifically: Three initial cluster center values ​​are randomly selected as the starting centers for each category. For the growth score index of each grid cell in the vegetation cover area, the distance between it and each cluster center is calculated, and it is assigned to the category of the nearest cluster center. For each category, calculate the mean of the growth score index of all grid cells assigned to that category, and use it as the new cluster center; Based on the updated cluster centers, the distance between each grid cell in the vegetation cover area and the updated cluster centers is calculated repeatedly according to the growth score index of each grid cell. The cells are then assigned to the category of the nearest cluster center until the cluster centers no longer change, thus completing the clustering process.

9. The method for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing according to claim 1, characterized in that, The method of setting a judgment threshold based on the clustering of dryland oat growth to achieve growth classification is as follows: Ultimately, all growth score indices were divided into: The clusters are obtained by sorting them in ascending order based on their cluster centers: ； in, , These represent the smallest, medium, and largest cluster centers of the growth score index, respectively. Calculate the dividing point of growth separately: ； ； in, This indicates the dividing line between weak and normal growth. This indicates the dividing line between normal and vigorous growth. when Less than or equal to At that time, it was determined that the dryland oats were in a weak growth condition; when Greater than and less than or equal to At that time, it was determined that the dryland oats were growing normally; when Greater than At that time, it was determined that the dryland oats were growing vigorously.

10. A system for monitoring the growth of dryland oats based on UAV hyperspectral remote sensing, characterized in that: The monitoring system is used to perform the monitoring method according to any one of claims 1-9, including: The data acquisition module is used to mark the boundaries of the target monitoring dryland oat planting area. Based on the marked boundaries, the planting area is divided into several identical rectangular grid units. The UAV is used to acquire the spectral images of each grid unit and preprocess the spectral images. The grid processing module is used to filter the reflectance bands of the preprocessed spectral image. For each grid cell, the average reflectance of all pixels in different bands is obtained to obtain the average reflectance data of the grid cell in different bands. All grid cells are classified according to the set band threshold and marked as bare soil area or vegetation cover area. The index construction module is used to dynamically calculate the adaptive parameters of the soil background based on the spectral images of all grid cells marked as bare soil areas. For grid cells marked as vegetated areas, multispectral coupling is used to construct vegetation spectral indices and water stress indices as indicators for monitoring the growth of dryland oats. The adaptive parameters of the soil background are used to correct the vegetation spectral indices and water stress indices corresponding to the grid cells in the vegetated areas. The growth scoring module is used to construct a growth scoring index based on the vegetation spectral index and water stress index corrected by the grid cells of each vegetation cover area, and to generate a growth scoring index after adaptive parameter correction of the soil background. The growth classification module is used to cluster the growth score index of each grid unit in the vegetation coverage area using the K-means clustering method, outputting the clustering status of dryland oats in the grid units of the vegetation coverage area, and setting a judgment threshold based on the clustering status of dryland oats to achieve growth classification.

Images