Optical-based methods and systems for monitoring the growth of Sichuan pepper

Abstract

This application relates to the field of growth monitoring, and discloses a method and system for monitoring the growth of Sichuan pepper based on optical means. The method includes: determining the spatial range of each Sichuan pepper planting plot in a target planting area; extracting the red-edge spectral features and near-infrared spectral features of Sichuan pepper from multispectral images of the target planting area during each period of the target phenological stage, based on the spatial range of the Sichuan pepper planting plots; and sequentially arranging the red-edge spectral features and near-infrared spectral features of each Sichuan pepper planting plot according to the imaging date of the multispectral images, thereby constructing a time-series chain of red-edge features and a time-series chain of near-infrared features for each Sichuan pepper planting plot. By extracting and analyzing the red-edge spectral features and near-infrared spectral features, accurate monitoring of the growth status of Sichuan pepper can be achieved. This method can reflect the growth status and health level of the plants in a timely manner, helping farmers make scientific decisions.

Description

Technical Field

[0001] This application relates to the field of growth monitoring, and in particular to a method and system for monitoring the growth of Sichuan pepper based on optical means. Background Technology

[0002] Climate change and environmental pressures pose more challenges to agriculture, such as extreme weather, soil degradation, and water shortages. Optical monitoring methods can help farmers respond to these changes in a timely manner, optimize management strategies, and improve stress resistance and adaptability by accurately monitoring plant status.

[0003] Currently, manual observation is often influenced by personal experience and judgment, which may lead to inconsistencies and subjectivity in monitoring results, making it difficult to guarantee the objectivity and accuracy of the data. Moreover, traditional methods usually require a lot of manpower and time for on-site inspections and recording, which is inefficient, especially in the case of large-scale planting, and cannot reflect the growth status of plants in real time.

[0004] In addition, traditional methods may not be able to fully obtain the multidimensional information required for plant growth, such as light and soil moisture, which will affect the comprehensive assessment of plant health status. Moreover, due to the need for regular inspections, traditional methods may not be able to detect abnormal growth or pest and disease risks in a timely manner, causing farmers to miss the best time for treatment, thus affecting yield and quality. Summary of the Invention

[0005] This application provides an optical method for monitoring the growth of Sichuan pepper, which at least partially solves the above-mentioned technical problems.

[0006] To achieve the above objectives, according to the first aspect of this application, a method for monitoring the growth of Sichuan pepper based on optical means is provided, comprising: Determine the spatial extent of each Sichuan pepper planting plot within the target planting area; Based on the spatial range of the Sichuan pepper planting area, the red edge spectral features and near-infrared spectral features of Sichuan pepper are extracted from the multispectral images of the target planting area during the target phenological period. The red edge spectral features and near-infrared spectral features of each pepper planting plot were sequentially arranged according to the imaging date of the multispectral images, and the corresponding red edge feature time series and near-infrared feature time series of each pepper planting plot were constructed. Based on the red-edge feature time series and near-infrared feature time series of each pepper planting plot, a comprehensive growth feature time series of each pepper planting plot is constructed; the comprehensive growth feature time series includes multiple growth feature nodes. Spatial clustering analysis was performed on multiple growth characteristic nodes of each comprehensive growth characteristic time series to obtain the clustering distribution results of each pepper planting plot; based on the clustering distribution results, the growth stages of each pepper planting plot were divided to obtain the growth status category of each pepper planting plot.

[0007] Optionally, the spatial extent of each Sichuan pepper planting plot within the target planting area can be determined, including: Acquire multispectral images of each target phenological period; Using the trained canopy segmentation neural network, the canopy of pepper trees was extracted from each multispectral image, and the initial canopy extraction results corresponding to each multispectral image were obtained. Based on the imaging geographic coordinates of multispectral images, the pixel coordinates of each preliminary screening canopy extraction result are transformed by map projection to obtain the actual distribution range corresponding to each preliminary screening canopy extraction result. Based on the actual distribution range corresponding to the tree canopy extraction results of each initial screening, the spatial range of each pepper planting plot is obtained.

[0008] Optionally, after performing map projection transformation on the pixel coordinates of each preliminary canopy extraction result based on the imaging geographic coordinates of multispectral imagery to obtain the actual distribution range corresponding to each preliminary canopy extraction result, the method further includes: Obtain non-vegetation cover data for the target planting area during the target phenological period; Based on non-vegetation cover data, bare ground masks corresponding to the imaging date and geographic coordinates of each multispectral image were constructed. According to the principle of spatiotemporal correspondence matching, each bare ground mask is matched with each multispectral image; Each bare ground mask was used to mask the initial tree canopy extraction results of the matched multispectral images to remove roads and water bodies that were mistakenly extracted as pepper trees in the initial tree canopy extraction results.

[0009] Optionally, based on the spatial extent of the Sichuan pepper planting area, the red-edge spectral features and near-infrared spectral features corresponding to Sichuan pepper are extracted from the multispectral images of the target planting area during the target phenological period, including: Obtain the band response variation characteristics of pepper planting plots; Based on the red edge band and visible light band in the band response change characteristics, the vegetation red edge index is determined as the red edge spectral characteristic of Sichuan pepper. From the remaining band response variation characteristics, the near-infrared band reflectance was extracted as the near-infrared spectral characteristics of Sichuan pepper. Among them, the band response change characteristics include the top M spectral bands with the largest reflectance difference between the budding and leaf-falling stages of the pepper planting plot in the same multispectral image data; M is greater than or equal to 3.

[0010] Optionally, spatial clustering analysis is performed on multiple growth characteristic nodes of each comprehensive growth characteristic time series to obtain the clustering distribution results of each pepper planting plot, including: Imaging date, red-edge spectral features, and near-infrared spectral features are used as clustering factors to perform three-dimensional spatial clustering of multiple growth feature nodes in the comprehensive growth feature time series, resulting in feature clusters. If the number of feature clusters is 1, the feature clusters are obtained as the clustering distribution results of the pepper planting plots; otherwise, each feature cluster is divided into high-activity clusters and low-activity clusters according to the mean of the red edge spectral features or the mean of the near-infrared spectral features. Multiple growth feature nodes in the high-activity clusters are re-clustered in three-dimensional space to obtain high-activity cluster branches of the high-activity clusters; Furthermore, multiple growth characteristic nodes in the low-activity clusters are re-clustered in three-dimensional space to obtain low-activity cluster branches of the low-activity clusters. The high-activity clusters, low-activity clusters, high-activity cluster branches, and low-activity cluster branches are used as the cluster distribution results of the pepper planting plots.

[0011] Optionally, based on the clustering distribution results, the growth stages of each Sichuan pepper planting plot are divided to obtain the growth status category of each Sichuan pepper planting plot, including: If the clustering results show that there is only one characteristic cluster, or that the number of high-activity cluster branches and the number of low-activity cluster branches are both less than or equal to 1, then the pepper planting plot is determined to be a normal growing plot. If not, the pepper planting area is determined to be an abnormal growth area.

[0012] Optionally, after classifying the growth stages of each Sichuan pepper planting plot based on the clustering distribution results to obtain the growth status category of each Sichuan pepper planting plot, the method further includes: If the clustering distribution results of the normally growing plots show whether there is only one characteristic cluster, whether the normally growing plots meet the preset red-edge feature threshold requirements, or whether there is a cluster branch in the clustering distribution results of the normally growing plots whose start and end imaging dates are located within the budding and harvesting periods of the target phenological period, respectively, then the phenological period type of the normally growing plots is determined to be full-cycle growth within the target phenological period. If not, then check whether the initial imaging date of the low-activity clusters of the normally growing plot is earlier than the initial imaging date of its high-activity clusters. If yes, then determine the phenological type of the normally growing plot as the late-stage fast-growing type within the target phenological period; if not, then determine the phenological type of the normally growing plot as the early-stage fast-growing type within the target phenological period.

[0013] Optionally, after classifying the growth stages of each Sichuan pepper planting plot based on the clustering distribution results to obtain the growth status category of each Sichuan pepper planting plot, the method further includes: Based on the first spectral data collected by the first monitoring point in the target planting area through the optical sensor during the first monitoring period, when the reflectance of a specific band in the first spectral data exceeds the first preset threshold, the first vegetation index of the first monitoring point during the first monitoring period is determined based on the first spectral data and the corresponding ambient light intensity data during the first monitoring period; wherein, the specific band is the band related to the chlorophyll content of pepper leaves. Based on the first vegetation index of the first monitoring point in the first monitoring period, combined with the historical growth data of the target planting area, the growth model of Sichuan pepper in the corresponding area is obtained, and the parameters of the growth model are fitted to obtain the corresponding first growth trend curve. Based on the first growth trend curve, the expected growth state of the first monitoring point in the second monitoring period after the first monitoring period is predicted to obtain the first predicted growth state; according to the first predicted growth state, the acquisition parameters of the optical sensor for the first monitoring point in the second monitoring period are adjusted, including the acquisition frequency and acquisition angle, and the second spectral data of the first monitoring point in the second monitoring period are obtained using the adjusted acquisition parameters. Based on the second spectral data, combined with the environmental temperature and humidity data during the second monitoring period, the health status of the pepper plants at the first monitoring point was analyzed to determine whether there was a risk of pests and diseases. Based on the pest and disease risk analysis obtained from the first monitoring point during the second monitoring period, the overall pest and disease occurrence probability distribution of the target planting area during the third monitoring period is predicted. The second monitoring period is earlier than the third monitoring period.

[0014] Optionally, based on the first spectral data and the corresponding ambient light intensity data within the first monitoring period, a first vegetation index is determined for the first monitoring point within the first monitoring period, including: Based on ambient light intensity data, the first spectral data is corrected for light intensity to obtain corrected spectral data. Based on the corrected spectral data, the reflectance values ​​of specific spectral bands are extracted; Based on the reflectance value of a specific band and the pre-set vegetation index calculation rules, combined with the correlation coefficient between growth rate and environmental meteorological factors, the first vegetation index of the first monitoring point in the first monitoring period is calculated.

[0015] According to a second aspect of this application, an optical-based system for monitoring the growth of Sichuan pepper is provided, comprising: The scope determination module is used to determine the spatial scope of each pepper planting plot in the target planting area; The feature extraction module is used to extract the red edge spectral features and near-infrared spectral features of Sichuan pepper from the multispectral images of the target planting area during the target phenological period, based on the spatial range of the Sichuan pepper planting plot. The feature arrangement module is used to sequentially arrange the red edge spectral features and near-infrared spectral features of each pepper planting plot according to the imaging date of the multispectral image, and construct the red edge feature time series chain and near-infrared feature time series chain of each pepper planting plot. The feature fusion module is used to construct the comprehensive growth feature time series of each pepper planting plot based on the red edge feature time series chain and the near-infrared feature time series chain of each pepper planting plot; the comprehensive growth feature time series includes multiple growth feature nodes; The growth clustering module is used to perform spatial clustering analysis on multiple growth feature nodes of each comprehensive growth feature time series to obtain the clustering distribution results of each pepper planting plot; based on the clustering distribution results, the growth stages of each pepper planting plot are divided to obtain the growth status category of each pepper planting plot.

[0016] In summary, the embodiments of this application, by extracting and analyzing red-edge spectral features and near-infrared spectral features, can achieve precise monitoring of the growth status of Sichuan pepper. This method can reflect the growth status and health level of the plants in a timely manner, helping farmers make scientific decisions. Moreover, by performing spatial cluster analysis on multiple growth feature nodes, the cluster distribution results of Sichuan pepper planting plots can be identified, and then the growth stages can be divided. This method can reveal the growth differences of different plots in the region, facilitating precise management. Furthermore, by considering environmental factors such as temperature and humidity, the spectral data is corrected and analyzed, making the monitoring results more reliable and applicable, and adaptable to Sichuan pepper planting under different climatic conditions.

[0017] Other features and advantages of this application will be described in detail in the following detailed description section. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the steps of the optical-based method for monitoring the growth of Sichuan pepper provided in an exemplary embodiment of this application; Figure 2 This is a schematic diagram of a pepper growth monitoring system based on optical means provided in an exemplary embodiment of this application; Explanation of reference numerals in the attached diagram: 1. Range determination module; 2. Feature extraction module; 3. Feature arrangement module; 4. Feature fusion module; 5. Growth clustering module. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained in the art based on the embodiments of this application without creative effort are within the protection scope of this application.

[0021] This application provides a method for monitoring the growth of Sichuan pepper based on optical methods. Please refer to [link / reference]. Figure 1 The optical-based method for monitoring the growth of Sichuan pepper provided in this application includes the following steps: S1. Determine the spatial range of each pepper planting plot in the target planting area; S2. Based on the spatial range of the pepper planting area, extract the red edge spectral features and near-infrared spectral features of pepper from the multispectral images of the target planting area during the target phenological period. S3. Sequentially arrange the red edge spectral features and near-infrared spectral features of each pepper planting plot according to the imaging date of the multispectral image, and construct the corresponding red edge feature time series chain and near-infrared feature time series chain for each pepper planting plot. S4. Based on the red-edge feature time series chain and near-infrared feature time series chain of each pepper planting plot, construct the comprehensive growth feature time series of each pepper planting plot; the comprehensive growth feature time series includes multiple growth feature nodes; S5. Perform spatial clustering analysis on multiple growth characteristic nodes of each comprehensive growth characteristic time series to obtain the clustering distribution results of each pepper planting plot; based on the clustering distribution results, divide the growth stage of each pepper planting plot to obtain the growth status category of each pepper planting plot.

[0022] It is important to note that determining the spatial extent of each Sichuan pepper planting plot within the target planting area is crucial. This step defines the monitoring area, ensuring that subsequent data extraction and analysis focus on specific Sichuan pepper planting plots. After determining the monitoring area, multispectral imagery is used to acquire the red-edge and near-infrared spectral characteristics of the Sichuan pepper planting plots at different phenological stages (such as germination, flowering, and fruiting). These spectral characteristics are essential for assessing plant health and growth status. The extracted red-edge and near-infrared spectral characteristics are then sequentially arranged according to the imaging date of the multispectral imagery. This step is necessary to construct a time series. This facilitates the observation and analysis of the growth trends of Sichuan pepper. Based on the aforementioned time series, a comprehensive growth characteristic time series for each Sichuan pepper planting plot is constructed. This time series contains multiple growth characteristic nodes, such as chlorophyll content and photosynthetic rate, which reflect the growth status of the plant at different time points. Spatial clustering analysis is performed on multiple growth characteristic nodes in these comprehensive growth characteristic time series to obtain the clustering distribution results of each Sichuan pepper planting plot. According to the clustering results, each planting plot can be divided into different growth stages to determine its growth status category, such as healthy, slow-growing, or vigorous-growing.

[0023] Specific example: Suppose there are three Sichuan pepper planting plots in a certain area, namely plot A, plot B, and plot C; the growth of Sichuan pepper in these plots is monitored using optical methods. First, GIS (Geographic Information System) technology is used to determine the boundaries of each plot; for example, plot A is located on the south slope, plot B is on the east slope, and plot C is on the west slope; in this way, they clearly define the spatial range of monitoring; multispectral images of the area in spring, summer, and autumn are acquired; in the images, they extract the red edge spectral and near-infrared spectral features of plots A, B, and C; the spectrum of plot A shows high red edge reflectance, which may indicate healthy leaf condition, while the red edge reflectance of plot B is low, suggesting possible growth problems.

[0024] The extracted spectral features were arranged chronologically. For example, the red edge feature data of plot A in spring was R1, in summer it was R2, and in autumn it was R3. Similarly, plots B and C were arranged sequentially to form a time series. Combining these time series data, a comprehensive growth characteristic of each plot was constructed. For example, the comprehensive growth characteristic time series of plot A showed that its chlorophyll content increased continuously over time, indicating good growth. The characteristics of plot B, however, showed fluctuations, possibly due to drought. Spatial cluster analysis was performed on the comprehensive growth characteristics. They found that the feature nodes of plot A clustered into one class, indicating that the plot was in good growth condition. Plot B was classified into another class, indicating slow growth. Plot C's condition was somewhere in between. Based on these clustering results, corresponding management and fertilization plans could be formulated to optimize the growth conditions of each plot.

[0025] In some embodiments, determining the spatial extent of each Sichuan pepper planting plot within the target planting area includes: Acquire multispectral images of each target phenological period; Using the trained canopy segmentation neural network, the canopy of pepper trees was extracted from each multispectral image, and the initial canopy extraction results corresponding to each multispectral image were obtained. Based on the imaging geographic coordinates of multispectral images, the pixel coordinates of each preliminary screening canopy extraction result are transformed by map projection to obtain the actual distribution range corresponding to each preliminary screening canopy extraction result. Based on the actual distribution range corresponding to the tree canopy extraction results of each initial screening, the spatial range of each pepper planting plot is obtained.

[0026] It should be noted that multispectral images were acquired at different phenological stages (such as germination, flowering, and fruiting). These images provide spectral information about the plants at different growth stages, helping to analyze the plant's health status and growth characteristics. A trained deep learning model (canopy segmentation neural network) was used to process the acquired multispectral images to extract the canopy of the Sichuan pepper trees. This process aims to identify the canopy regions of the Sichuan pepper trees in the images, thereby eliminating background and other interfering information and focusing on the growth of the Sichuan pepper trees. After canopy segmentation, a preliminary canopy extraction result is obtained, showing the identified canopy regions of the Sichuan pepper trees in each multispectral image. This step is the basis for obtaining accurate block boundaries. Based on the geographic coordinates of the multispectral images, the pixel coordinates in the preliminary screening results are converted into actual geographic locations. This step uses map projection technology to map the canopy information extracted from the images to a real-world coordinate system, allowing the extraction results to be compared with the actual situation. Combining the canopy extraction results after map projection transformation, the actual spatial range of each Sichuan pepper planting block can be determined. This result provides important basic data for subsequent growth monitoring and management decisions.

[0027] Specific example: Suppose there are multiple Sichuan pepper planting plots in a mountainous area, and the goal is to determine the spatial extent of these plots using optical methods. Multispectral images of these plots were captured using drones in spring and summer. The spring images help analyze the budding status of the Sichuan pepper trees, while the summer images provide information on flowering and fruit growth. A pre-trained neural network model is used to process these multispectral images. For example, when given the spring multispectral images, the network successfully identifies and marks the canopies of the Sichuan pepper trees. After processing, preliminary canopy extraction results are obtained. For instance, the canopy extraction results for plot A in the spring images show approximately 200 Sichuan pepper trees, while the canopy extraction results for plot B show approximately 150 Sichuan pepper trees.

[0028] Based on the geographic coordinates of the image, the extracted canopy pixel coordinates are transformed using map projection. For example, the canopy pixels of plot A have coordinates of (300, 450) in the image. After transformation, they are mapped to the actual geographic location as (30.1234°N, 102.5678°E). This transformation allows the extracted results to be accurately located on the map. Combining all the extracted canopy results, the actual spatial extent of plots A and B can be plotted. After analysis, the total area of ​​plot A is 5 mu (approximately 0.33 hectares), while the total area of ​​plot B is 3 mu (approximately 0.2 hectares). These data provide important basis for subsequent growth monitoring, fertilization, and irrigation planning.

[0029] In some embodiments, after performing map projection transformation on the pixel coordinates of each preliminary canopy extraction result based on the imaging geographic coordinates of multispectral imagery to obtain the actual distribution range corresponding to each preliminary canopy extraction result, the method further includes: Obtain non-vegetation cover data for the target planting area during the target phenological period; Based on non-vegetation cover data, bare ground masks corresponding to the imaging date and geographic coordinates of each multispectral image were constructed. According to the principle of spatiotemporal correspondence matching, each bare ground mask is matched with each multispectral image; Each bare ground mask was used to mask the initial tree canopy extraction results of the matched multispectral images to remove roads and water bodies that were mistakenly extracted as pepper trees in the initial tree canopy extraction results.

[0030] It should be noted that non-vegetation cover data of the target planting area during a specific phenological period is collected. This data typically includes information on bare land, roads, and water bodies, which can be obtained through satellite imagery or other remote sensing technologies. Based on the obtained non-vegetation cover data, a bare land mask is constructed for each multispectral image. This mask is a binary image, where bare land is marked as one and vegetation is marked as zero, thus clearly identifying non-vegetated areas. Following the spatiotemporal correspondence matching principle, each bare land mask is matched with its corresponding multispectral image. This means that only images and masks taken at the same time and geographical location are combined to ensure data consistency. The initial canopy extraction results of the multispectral images matched with each bare land mask are then masked. This step removes areas that are incorrectly identified as Sichuan pepper trees, such as roads and water bodies. Through masking, a more accurate distribution of Sichuan pepper tree canopies can be obtained.

[0031] Specific example: Suppose a county has a large area of ​​Sichuan pepper cultivation, and the goal is to improve the extraction quality of Sichuan pepper tree data using the methods described above. Satellite imagery was used to collect non-vegetation cover data for the county in summer. They discovered several main roads and some water bodies within the cultivation area, which needed to be removed in subsequent analysis. Based on this non-vegetation data, a bare ground mask was constructed for the summer multispectral imagery. For example, after processing, roads and water bodies in the image were marked as "1," while other areas (such as Sichuan pepper trees and other plants) were marked as "0." This resulted in a mask that clearly distinguishes non-vegetation areas. This bare ground mask was then matched with the summer multispectral imagery. Because the imagery and mask were acquired simultaneously in the same area, they corresponded perfectly, ensuring the accuracy of the analysis. This bare ground mask was then used to mask the previously extracted preliminary canopy results. They found that some canopies in the preliminary results were actually located near roads or even on water bodies. These areas were removed after masking, leaving only the more accurate locations of the Sichuan pepper tree canopies.

[0032] In some embodiments, based on the spatial extent of the Sichuan pepper planting area, the red-edge spectral features and near-infrared spectral features corresponding to Sichuan pepper are extracted from multispectral images of the target planting area during each stage of the target phenological period, including: Obtain the band response variation characteristics of pepper planting plots; Based on the red edge band and visible light band in the band response change characteristics, the vegetation red edge index is determined as the red edge spectral characteristic of Sichuan pepper. From the remaining band response variation characteristics, the near-infrared band reflectance was extracted as the near-infrared spectral characteristics of Sichuan pepper. Among them, the band response change characteristics include the top M spectral bands with the largest reflectance difference between the budding and leaf-falling stages of the pepper planting plot in the same multispectral image data; M is greater than or equal to 3.

[0033] It should be noted that multispectral imagery from multiple periods was used to analyze the reflectance changes of Sichuan pepper plantations at different growth stages (such as budding and leaf fall). These changes reflect the physiological status of plants at different growth stages. During the analysis, the response changes in the red-edge band and the visible light band were analyzed to calculate the vegetation red-edge index. The red-edge band refers to the band between the visible and near-infrared spectra, which is crucial for distinguishing different types of vegetation and reflecting plant health and growth status. In addition to red-edge spectral characteristics, reflectance in the near-infrared band was extracted from the response changes in the remaining bands. The near-infrared band is a good indicator of plant water content and growth status, thus providing important information about Sichuan pepper tree growth. The reflectance differences of Sichuan pepper plantations during the budding and leaf fall stages in the same multispectral image were compared, and the top M spectral bands with the largest reflectance differences were selected for analysis. Here, M is greater than or equal to 3, meaning that at least three bands need to be considered to ensure the representativeness of the extracted data.

[0034] Specific example: Suppose a region has multiple orchards in a Sichuan pepper-growing area, and the goal is to monitor the growth status of these orchards using multispectral imagery. Multispectral images taken in spring (budding stage) and autumn (leaf fall) were selected. These images were analyzed, recording the changes in reflectance of the Sichuan pepper trees during these two periods. They found that the reflectance of the Sichuan pepper trees was generally higher during the budding stage than during the leaf fall stage, particularly noticeable in the red-edge and near-infrared bands. By comparing the reflectance of different bands, the changes in the red-edge band and visible light bands (such as the red and blue bands) were identified. The vegetation red-edge index was calculated, indicating that the Sichuan pepper trees are healthy and vigorous during the budding stage, but weaker during the leaf fall stage. Changes in this index can help them assess the growth status of the Sichuan pepper trees.

[0035] They extracted near-infrared reflectance data related to the growth of Sichuan pepper from changes in band response. They found that the reflectance of the near-infrared band was as high as 0.7 during the budding stage, but decreased to 0.4 during the leaf fall stage. This change suggests that there is a significant difference in the water content of Sichuan pepper trees during the two stages. They compared the reflectance of Sichuan pepper planting plots in the same multispectral image during the budding and leaf fall stages and identified the top three bands with the largest reflectance differences. These bands correspond to the red edge, near-infrared, and green bands, respectively. The combination of these three provides a reliable data basis for subsequent monitoring of Sichuan pepper.

[0036] In some embodiments, spatial clustering analysis is performed on multiple growth characteristic nodes of each comprehensive growth characteristic time series to obtain the clustering distribution results of each Sichuan pepper planting plot, including: Imaging date, red-edge spectral features, and near-infrared spectral features are used as clustering factors to perform three-dimensional spatial clustering of multiple growth feature nodes in the comprehensive growth feature time series, resulting in feature clusters. If the number of feature clusters is 1, the feature clusters are obtained as the clustering distribution results of the pepper planting plots; otherwise, each feature cluster is divided into high-activity clusters and low-activity clusters according to the mean of the red edge spectral features or the mean of the near-infrared spectral features. Multiple growth feature nodes in the high-activity clusters are re-clustered in three-dimensional space to obtain high-activity cluster branches of the high-activity clusters; Furthermore, multiple growth characteristic nodes in the low-activity clusters are re-clustered in three-dimensional space to obtain low-activity cluster branches of the low-activity clusters. The high-activity clusters, low-activity clusters, high-activity cluster branches, and low-activity cluster branches are used as the cluster distribution results of the pepper planting plots.

[0037] It should be noted that imaging date, red-edge spectral features, and near-infrared spectral features are used as clustering factors. This means that during cluster analysis, these factors will be used to determine the growth status of Sichuan pepper planting plots under different times and different spectral characteristics. Based on these clustering factors, multiple growth feature nodes in the comprehensive growth feature time series are clustered in three dimensions. This step will divide different growth feature nodes into different clusters according to similarity to identify regions with similar growth states. After clustering is completed, the number of clusters will be detected. If only one cluster is formed, it means that all samples are similar in growth characteristics. In this case, the cluster can be directly used as the clustering distribution result of the Sichuan pepper planting plots. If multiple clusters are formed... Clusters require further analysis. If multiple clusters exist, they will be divided into high-activity clusters and low-activity clusters based on the mean values ​​of red-edge spectral features or near-infrared spectral features. High-activity clusters indicate good growth, while low-activity clusters indicate poor growth. Multiple growth characteristic nodes within both high-activity and low-activity clusters will be further clustered in three dimensions. This will allow for further subdivision of regions with similar growth characteristics, resulting in high-activity and low-activity cluster branches. These high-activity clusters, low-activity clusters, high-activity cluster branches, and low-activity cluster branches will collectively constitute the cluster distribution results of the Sichuan pepper planting area, providing a basis for subsequent management decisions.

[0038] Specific example: Suppose a pepper-growing area in a certain region is divided into several small plots, and the goal is to use multidimensional growth characteristics for cluster analysis to understand the growth status of different plots. Multispectral image data of each plot were collected at different times (e.g., spring, summer, and autumn), and red-edge spectral features and near-infrared spectral features were extracted. The imaging date, red-edge spectral features, and near-infrared spectral features were selected as clustering factors to perform three-dimensional spatial clustering of the growth characteristic nodes of each pepper-growing plot. Through analysis, they found that these plots were divided into three clusters, one of which showed a healthy growth status, while the other two showed a weaker growth status.

[0039] When detecting the number of clusters, three clusters were found. Since there was more than one cluster, they decided to divide these three clusters based on the mean of the red-edge spectral features. After analysis, it was found that one cluster had a high mean of red-edge spectral features, indicating that the pepper trees in this area were growing well, and therefore it was classified as a high-vitality cluster. The other two clusters had low mean values, indicating poor growth, and therefore they were classified as low-vitality clusters. For the high-vitality cluster, the growth feature nodes were further re-clustered, and it was found that they could be divided into two high-vitality cluster branches, corresponding to areas with better and better growth conditions, respectively. The low-vitality cluster was re-clustered, resulting in two low-vitality cluster branches, corresponding to areas with poor and very poor growth conditions, respectively.

[0040] In some embodiments, based on the clustering distribution results, the growth stages of each Sichuan pepper planting plot are divided to obtain the growth status category of each Sichuan pepper planting plot, including: If the clustering results show that there is only one characteristic cluster, or that the number of high-activity cluster branches and the number of low-activity cluster branches are both less than or equal to 1, then the pepper planting plot is determined to be a normal growing plot. If not, the pepper planting area is determined to be an abnormal growth area.

[0041] It should be noted that the clustering results are checked to see if there is only one characteristic cluster. Characteristic clusters represent areas with similar growth characteristics. If there is only one cluster, it means that all plots are relatively consistent in their growth status, indicating that they are all in a normal growth state. If there are multiple clusters in the clustering distribution, the number of high-activity and low-activity cluster branches is further checked. If the number of both branches is less than or equal to 1, this also indicates that the overall growth status is normal, because it means that there is only one high-activity area (or none) and one low-activity area (or none), and there is no obvious growth problem overall. If the clustering results show multiple characteristic clusters, or the number of high-activity and low-activity branches is greater than 1, this usually means that the growth status of different plots varies greatly, which may indicate that some plots have abnormal growth. Through the above judgment, the pepper planting plots can be divided into two categories: normal growth plots and abnormal growth plots. This classification is crucial for agricultural management and the formulation of follow-up measures.

[0042] Specific example: An analysis is being conducted on a Sichuan pepper planting area, which is subdivided into multiple plots. Cluster analysis is being performed to determine the growth status of each plot. Growth characteristic data for each plot are collected, and after cluster analysis, the following cluster distribution results are obtained. When analyzing the cluster distribution results, the number of clusters is first checked. It is found that all plots merge into one cluster, indicating that these plots are very similar in growth characteristics and have consistent growth status. Therefore, it is determined that the Sichuan pepper planting plots in this area are all growing normally. Then, suppose the cluster analysis results for another plot show two clusters, one high-activity cluster and the other low-activity cluster. At this point, the high-activity cluster will be further examined... The analysis included the number of high-activity and low-activity clusters. They found one high-activity cluster and one low-activity cluster. This indicated that while there were high-activity and low-activity areas, the number of branches for each type did not exceed one. Therefore, the overall pepper-growing plots in this area could still be considered to be in a normal growth state. However, if the same cluster analysis revealed three clusters, two of which were high-activity and the other low-activity, it meant that the growth status of different plots differed significantly, and there might be multiple areas with abnormal growth. Therefore, this area would be marked as an abnormal growth plot, and a detailed investigation would be recommended to identify the causes of the abnormal growth, such as soil problems, pests and diseases, or other environmental factors.

[0043] In some embodiments, after classifying the growth stages of each Sichuan pepper planting plot based on clustering distribution results to obtain the growth status category of each Sichuan pepper planting plot, the method further includes: If the clustering distribution results of the normally growing plots show whether there is only one characteristic cluster, whether the normally growing plots meet the preset red-edge feature threshold requirements, or whether there is a cluster branch in the clustering distribution results of the normally growing plots whose start and end imaging dates are located within the budding and harvesting periods of the target phenological period, respectively, then the phenological period type of the normally growing plots is determined to be full-cycle growth within the target phenological period. If not, then check whether the initial imaging date of the low-activity clusters of the normally growing plot is earlier than the initial imaging date of its high-activity clusters. If yes, then determine the phenological type of the normally growing plot as the late-stage fast-growing type within the target phenological period; if not, then determine the phenological type of the normally growing plot as the early-stage fast-growing type within the target phenological period.

[0044] It should be noted that the clustering distribution results of normally growing plots are checked to confirm whether there is only one characteristic cluster. If only one cluster exists, it means that the growth in this area is relatively uniform. The pre-defined red-edge feature threshold is then assessed. The red-edge feature is usually related to the health of the plants; if the threshold is met, it indicates that the plants are growing well. The clustering distribution results of normally growing plots also need to be checked to determine if the start and end imaging dates of any cluster branch fall within the budding and harvesting periods of the target phenological period, respectively. If both conditions are met, the phenological type of the plot is determined to be "growth throughout the entire target phenological period," indicating that the entire growth cycle is within the normal range. If the above conditions are not fully met, further checks will be conducted to determine whether the initial imaging date of low-activity clusters in normally growing plots is earlier than that of high-activity clusters. If the initial imaging date of low-activity clusters is earlier than that of high-activity clusters, the plot can be identified as "late-stage fast-growing type within the target phenological period," indicating that the growth rate in this area significantly accelerates in the later stages of the phenological period. If the initial imaging date of low-activity clusters is later than or equal to that of high-activity clusters, it is identified as "early-stage fast-growing type within the target phenological period," indicating that the growth rate is faster in the early stages of the phenological period. Through these steps, a deeper understanding of the growth cycle and phenological period type of Sichuan pepper planting plots can be achieved, thereby providing a scientific basis for agricultural management.

[0045] Specific examples: Suppose there is a Sichuan pepper planting area. After preliminary cluster analysis, it was found that the normally growing plots in this area formed only one characteristic cluster, and the red-edge feature data of this plot met the preset threshold requirements. Next, they checked the imaging dates of this cluster and found that its start imaging date was March 1st and the end imaging date was August 31st, which exactly included the budding period (March to May) and harvesting period (July to August) of the target phenological period. Therefore, it was determined that the phenological period type of this plot was full-cycle growth within the target phenological period, indicating that the overall growth cycle was in good condition. Another example is the analysis of another normally growing plot, which showed that the cluster distribution of this plot had two clusters, among which... One cluster is high-activity cluster, and the other is low-activity cluster. Evaluation revealed that the low-activity cluster's initial imaging date was April 1st, while the high-activity cluster's was May 1st. Because the low-activity cluster's start date was earlier than the high-activity cluster's, the phenological type of this plot was determined to be a late-stage fast-growing type within the target phenological period, indicating that the plot's growth rate accelerated in the later stages of the phenological period. Finally, assuming another plot had a low-activity cluster's initial imaging date of May 15th, while the high-activity cluster's was May 10th, in this case, because the low-activity cluster's date was later than the high-activity cluster's date, the plot's phenological type would be considered an early-stage fast-growing type within the target phenological period.

[0046] In some embodiments, after classifying the growth stages of each Sichuan pepper planting plot based on clustering distribution results to obtain the growth status category of each Sichuan pepper planting plot, the method further includes: Based on the first spectral data collected by the first monitoring point in the target planting area through the optical sensor during the first monitoring period, when the reflectance of a specific band in the first spectral data exceeds the first preset threshold, the first vegetation index of the first monitoring point during the first monitoring period is determined based on the first spectral data and the corresponding ambient light intensity data during the first monitoring period; wherein, the specific band is the band related to the chlorophyll content of pepper leaves. Based on the first vegetation index of the first monitoring point in the first monitoring period, combined with the historical growth data of the target planting area, the growth model of Sichuan pepper in the corresponding area is obtained, and the parameters of the growth model are fitted to obtain the corresponding first growth trend curve. Based on the first growth trend curve, the expected growth state of the first monitoring point in the second monitoring period after the first monitoring period is predicted to obtain the first predicted growth state; according to the first predicted growth state, the acquisition parameters of the optical sensor for the first monitoring point in the second monitoring period are adjusted, including the acquisition frequency and acquisition angle, and the second spectral data of the first monitoring point in the second monitoring period are obtained using the adjusted acquisition parameters. Based on the second spectral data, combined with the environmental temperature and humidity data during the second monitoring period, the health status of the pepper plants at the first monitoring point was analyzed to determine whether there was a risk of pests and diseases. Based on the pest and disease risk analysis obtained from the first monitoring point during the second monitoring period, the overall pest and disease occurrence probability distribution of the target planting area during the third monitoring period is predicted. The second monitoring period is earlier than the third monitoring period.

[0047] It should be noted that at the first monitoring point in the target planting area, spectral data was collected using optical sensors during the first monitoring period. Measuring reflectance in specific wavelength bands is crucial, especially those bands related to chlorophyll content in pepper leaves. When the reflectance in these bands exceeds a set threshold, it indicates good plant growth. Using the collected first spectral data and the corresponding ambient light intensity, the vegetation index for that monitoring point during the first monitoring period was calculated. This index helps determine the plant's health status. The vegetation index at the first monitoring point was combined with historical growth data from the target planting area to construct a corresponding pepper growth model. After parameter fitting, this model will generate a first growth trend curve, showing the expected growth of pepper in the region. The system calculates the growth status of the first monitoring point within the second monitoring period based on the first growth trend curve, and obtains the first predicted growth status. Based on the first predicted growth status, the acquisition parameters of the optical sensor within the second monitoring period, such as acquisition frequency and acquisition angle, are adjusted to optimize data collection. Using these adjusted parameters, the second spectral data of the first monitoring point within the second monitoring period is obtained. Based on the second spectral data and the environmental temperature and humidity data within the second monitoring period, the health status of the Sichuan pepper plants at the first monitoring point is analyzed to determine if there is a risk of pests or diseases. Based on the pest and disease risk information obtained during the second monitoring period, the overall probability distribution of pests and diseases that may occur in the target planting area during the third monitoring period is predicted.

[0048] Specific example: Imagine a Sichuan pepper plantation located in Northwest China. To ensure the healthy growth of the peppers, the farmer decides to use the above-mentioned monitoring and management methods. In the first monitoring period, the farmer uses an optical sensor to collect spectral data at a monitoring point in the plantation. Analysis reveals that the reflectance in certain wavelengths (such as 660nm and 740nm) exceeds a preset threshold, indicating that the leaves of these pepper trees have a high chlorophyll content and are growing well. Therefore, the farmer calculates the first vegetation index for this monitoring point, such as NDVI (Normalized Difference Vegetation Index). Next, the farmer combines this vegetation index with historical growth data from the past few years to establish a Sichuan pepper growth model. After parameter fitting, the model obtains a first growth trend curve that reflects the growth of Sichuan pepper in the region, showing the growth trend of the pepper in the coming months. The situation is expected to continue to improve. With this information, the farmer can predict that the pepper trees will continue to grow healthily in the second monitoring cycle. Therefore, he decided to increase the frequency of spectral data collection in the second monitoring cycle to better capture information about changes in plant growth and to adjust the sensor's collection angle to ensure coverage of more leaves. In the second monitoring cycle, the farmer obtained second spectral data through the new settings and analyzed it in conjunction with the temperature and humidity data at the time. The results showed that although the pepper trees were generally healthy, there were slight signs of pests and diseases in a certain area of ​​the leaves, indicating a decline in plant health. Based on this finding, the farmer took corresponding prevention and control measures before the third monitoring cycle and used the analyzed pest and disease risk information to predict the probability distribution of pests and diseases in the entire planting area in the next cycle, so as to make corresponding management and prevention preparations in advance.

[0049] In some embodiments, based on first spectral data and corresponding ambient light intensity data within a first monitoring period, a first vegetation index for a first monitoring point within the first monitoring period is determined, including: Based on ambient light intensity data, the first spectral data is corrected for light intensity to obtain corrected spectral data. Based on the corrected spectral data, the reflectance values ​​of specific spectral bands are extracted; Based on the reflectance value of a specific band and the pre-set vegetation index calculation rules, combined with the correlation coefficient between growth rate and environmental meteorological factors, the first vegetation index of the first monitoring point in the first monitoring period is calculated.

[0050] It should be noted that the spectral data collected during the first monitoring period may be affected by changes in ambient light intensity, thus requiring correction. By analyzing the ambient light intensity data, the original spectral data can be adjusted to eliminate interference from light variations, resulting in more accurate spectral data. After correction, the next step is to extract reflectance values ​​for specific wavelengths from the corrected spectral data. These specific wavelengths are generally related to plant growth and chlorophyll content, such as red and near-infrared light bands. By extracting the reflectance of these wavelengths, the health status of plants can be better assessed. With the reflectance values ​​of specific wavelengths, combined with pre-set vegetation index calculation rules, the first vegetation index for the first monitoring point can be calculated. In addition, the correlation coefficient between growth rate and environmental meteorological factors will be considered to better reflect the plant growth under specific environmental conditions.

[0051] Specific example: Suppose there's a Sichuan pepper plantation on a hillside, and environmental monitoring is monitoring the growth of the pepper trees to ensure their healthy development. In the first monitoring cycle, they collect spectral data using optical sensors and record the ambient light intensity of the area. For instance, on the day of data collection, the sunlight was intense, reaching 1000 watts per square meter. To eliminate the influence of light variations on the spectral data, they correct the original spectral data. For example, if the reflectance of a specific wavelength (such as the red band) is 0.3 during peak sunlight, after light intensity correction, the corrected reflectance might become 0.25, thus accurately reflecting the plant's health. Specific parameters are then extracted from the corrected spectral data. The reflectance values ​​for specific wavelengths, such as 0.25 for the red band and 0.6 for the near-infrared band, are used to calculate vegetation indices using pre-defined rules. For example, they might use the NDVI (Normalized Difference Vegetation Index) to determine plant growth status. In this process, the correlation between other environmental factors, such as temperature, humidity, and rainfall, and growth rate is considered. For instance, if temperatures are moderate and there is adequate rainfall in recent days, resulting in faster plant growth, the correlation coefficient might be higher. By combining these factors, they ultimately calculate the first vegetation index for the first monitoring point, showing good plant growth and a relatively high health index.

[0052] Reference Figure 2 The second embodiment of the present invention provides a pepper growth monitoring system based on optical means, comprising: Module 1, which is used to determine the spatial range of each pepper planting plot in the target planting area; Feature extraction module 2 is used to extract the red edge spectral features and near-infrared spectral features of Sichuan pepper from the multispectral images of the target planting area during the target phenological period, based on the spatial range of the Sichuan pepper planting plot. Feature arrangement module 3 is used to sequentially arrange the red edge spectral features and near-infrared spectral features of each pepper planting plot according to the imaging date of the multispectral image, and construct the red edge feature time series chain and near-infrared feature time series chain of each pepper planting plot accordingly. Feature fusion module 4 is used to construct the comprehensive growth feature time series of each pepper planting plot based on the red edge feature time series chain and near-infrared feature time series chain of each pepper planting plot; the comprehensive growth feature time series includes multiple growth feature nodes; The growth clustering module 5 is used to perform spatial clustering analysis on multiple growth feature nodes of each comprehensive growth feature time series to obtain the clustering distribution results of each pepper planting plot; based on the clustering distribution results, the growth stages of each pepper planting plot are divided to obtain the growth status category of each pepper planting plot.

[0053] It should be noted that the optical-based pepper growth monitoring system provided in this embodiment of the invention is used to execute all the process steps of the optical-based pepper growth monitoring method in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0054] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0055] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0056] The embodiments, implementation methods, and related technical features of this application can be combined and substituted for each other without conflict.

[0057] The above are merely preferred embodiments of this application and are not intended to limit this application in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of this application without departing from the scope of the technical solution of this application shall still fall within the scope of the technical solution of this application.

Claims

1. A method for monitoring the growth of Sichuan pepper based on optical means, characterized in that, include: Determine the spatial extent of each Sichuan pepper planting plot within the target planting area; Based on the spatial range of the Sichuan pepper planting area, the red edge spectral features and near-infrared spectral features of Sichuan pepper are extracted from the multispectral images of the target planting area during the target phenological period. The red edge spectral features and near-infrared spectral features of each pepper planting plot were sequentially arranged according to the imaging date of the multispectral images, and the corresponding red edge feature time series and near-infrared feature time series of each pepper planting plot were constructed. Based on the red-edge feature time series and near-infrared feature time series of each pepper planting plot, a comprehensive growth feature time series of each pepper planting plot is constructed; the comprehensive growth feature time series includes multiple growth feature nodes. Spatial clustering analysis was performed on multiple growth characteristic nodes of each comprehensive growth characteristic time series to obtain the clustering distribution results of each pepper planting plot; based on the clustering distribution results, the growth stages of each pepper planting plot were divided to obtain the growth status category of each pepper planting plot.

2. The method according to claim 1, characterized in that, Determine the spatial extent of each Sichuan pepper planting plot within the target planting area, including: Acquire multispectral images of each target phenological period; Using the trained canopy segmentation neural network, the canopy of pepper trees was extracted from each multispectral image, and the initial canopy extraction results corresponding to each multispectral image were obtained. Based on the imaging geographic coordinates of multispectral images, the pixel coordinates of each preliminary screening canopy extraction result are transformed by map projection to obtain the actual distribution range corresponding to each preliminary screening canopy extraction result. Based on the actual distribution range corresponding to the tree canopy extraction results of each initial screening, the spatial range of each pepper planting plot is obtained.

3. The method according to claim 2, characterized in that, After performing map projection transformation on the pixel coordinates of each preliminary canopy extraction result based on the imaging geographic coordinates of multispectral imagery to obtain the actual distribution range corresponding to each preliminary canopy extraction result, the method further includes: Obtain non-vegetation cover data for the target planting area during the target phenological period; Based on non-vegetation cover data, bare ground masks corresponding to the imaging date and geographic coordinates of each multispectral image were constructed. According to the principle of spatiotemporal correspondence matching, each bare ground mask is matched with each multispectral image; Each bare ground mask was used to mask the initial tree canopy extraction results of the matched multispectral images to remove roads and water bodies that were mistakenly extracted as pepper trees in the initial tree canopy extraction results.

4. The method according to claim 3, characterized in that, Based on the spatial extent of the Sichuan pepper planting area, the red-edge spectral features and near-infrared spectral features corresponding to Sichuan pepper were extracted from multispectral images of the target planting area during the target phenological period, including: Obtain the band response variation characteristics of pepper planting plots; Based on the red edge band and visible light band in the band response change characteristics, the vegetation red edge index is determined as the red edge spectral characteristic of Sichuan pepper. From the remaining band response variation characteristics, the near-infrared band reflectance was extracted as the near-infrared spectral characteristics of Sichuan pepper. Among them, the band response change characteristics include the top M spectral bands with the largest reflectance difference between the budding and leaf-falling stages of the pepper planting plot in the same multispectral image data; M is greater than or equal to 3.

5. The method according to claim 4, characterized in that, Spatial clustering analysis was performed on multiple growth characteristic nodes of each comprehensive growth characteristic time series to obtain the clustering distribution results of each pepper planting plot, including: Imaging date, red-edge spectral features, and near-infrared spectral features are used as clustering factors to perform three-dimensional spatial clustering of multiple growth feature nodes in the comprehensive growth feature time series, resulting in feature clusters. If the number of feature clusters is 1, the feature clusters are obtained as the clustering distribution results of the pepper planting plots; otherwise, each feature cluster is divided into high-activity clusters and low-activity clusters according to the mean of the red edge spectral features or the mean of the near-infrared spectral features. Multiple growth feature nodes in the high-activity clusters are re-clustered in three-dimensional space to obtain high-activity cluster branches; and, Multiple growth characteristic nodes in the low-activity clusters were re-clustered in three-dimensional space to obtain low-activity cluster branches. The high-activity clusters, low-activity clusters, high-activity cluster branches, and low-activity cluster branches were used as the cluster distribution results of the pepper planting plots.

6. The method according to claim 5, characterized in that, Based on the clustering distribution results, the growth stages of each Sichuan pepper planting plot were divided, resulting in the growth status categories of each plot, including: If the clustering results show that there is only one characteristic cluster, or that the number of high-activity cluster branches and the number of low-activity cluster branches are both less than or equal to 1, then the pepper planting plot is determined to be a normal growing plot. If not, the pepper planting area is determined to be an abnormal growth area.

7. The method according to claim 6, characterized in that, After classifying the growth stages of each Sichuan pepper planting plot based on the clustering distribution results to obtain the growth status category of each Sichuan pepper planting plot, the method further includes: The clustering results of normally growing plots are checked to determine if there is only one characteristic cluster, whether the normally growing plots meet the preset red-edge feature threshold, or whether the clustering results of normally growing plots have a cluster branch whose start and end imaging dates are located within the budding and harvesting periods of the target phenological period, respectively. If so, the phenological period type of the normally growing plots is determined to be full-cycle growth within the target phenological period; otherwise, the phenological period type is determined to be... If the initial imaging date of the low-activity clusters of the normally growing plot is earlier than the initial imaging date of its high-activity clusters, then the phenological type of the normally growing plot is determined to be the late-stage fast-growing type within the target phenological period; otherwise, the phenological type of the normally growing plot is determined to be the early-stage fast-growing type within the target phenological period.

8. The method according to claim 7, characterized in that, After classifying the growth stages of each Sichuan pepper planting plot based on the clustering distribution results to obtain the growth status category of each Sichuan pepper planting plot, the method further includes: Based on the first spectral data collected by the first monitoring point in the target planting area through the optical sensor during the first monitoring period, when the reflectance of a specific band in the first spectral data exceeds the first preset threshold, the first vegetation index of the first monitoring point during the first monitoring period is determined based on the first spectral data and the corresponding ambient light intensity data during the first monitoring period; wherein, the specific band is the band related to the chlorophyll content of pepper leaves. Based on the first vegetation index of the first monitoring point in the first monitoring period, combined with the historical growth data of the target planting area, the growth model of Sichuan pepper in the corresponding area is obtained, and the parameters of the growth model are fitted to obtain the corresponding first growth trend curve. Based on the first growth trend curve, the expected growth state of the first monitoring point in the second monitoring period after the first monitoring period is predicted to obtain the first predicted growth state; according to the first predicted growth state, the acquisition parameters of the optical sensor for the first monitoring point in the second monitoring period are adjusted, including the acquisition frequency and acquisition angle, and the second spectral data of the first monitoring point in the second monitoring period are obtained using the adjusted acquisition parameters. Based on the second spectral data, combined with the environmental temperature and humidity data during the second monitoring period, the health status of the pepper plants at the first monitoring point was analyzed to determine whether there was a risk of pests and diseases. Based on the pest and disease risk analysis obtained from the first monitoring point during the second monitoring period, the overall pest and disease occurrence probability distribution of the target planting area during the third monitoring period is predicted. The second monitoring period is earlier than the third monitoring period.

9. The method according to claim 8, characterized in that, Based on the first spectral data and the corresponding ambient light intensity data within the first monitoring period, the first vegetation index of the first monitoring point within the first monitoring period is determined, including: Based on ambient light intensity data, the first spectral data is corrected for light intensity to obtain corrected spectral data. Based on the corrected spectral data, the reflectance values ​​of specific spectral bands are extracted; Based on the reflectance value of a specific band and the pre-set vegetation index calculation rules, combined with the correlation coefficient between growth rate and environmental meteorological factors, the first vegetation index of the first monitoring point in the first monitoring period is calculated.

10. A pepper growth monitoring system based on optical methods, applicable to the method described in any one of claims 1 to 9, characterized in that, include: The scope determination module is used to determine the spatial scope of each pepper planting plot in the target planting area; The feature extraction module is used to extract the red edge spectral features and near-infrared spectral features of Sichuan pepper from the multispectral images of the target planting area during the target phenological period, based on the spatial range of the Sichuan pepper planting plot. The feature arrangement module is used to sequentially arrange the red edge spectral features and near-infrared spectral features of each pepper planting plot according to the imaging date of the multispectral image, and construct the red edge feature time series chain and near-infrared feature time series chain of each pepper planting plot. The feature fusion module is used to construct the comprehensive growth feature time series of each pepper planting plot based on the red edge feature time series chain and the near-infrared feature time series chain of each pepper planting plot; the comprehensive growth feature time series includes multiple growth feature nodes; The growth clustering module is used to perform spatial clustering analysis on multiple growth feature nodes of each comprehensive growth feature time series to obtain the clustering distribution results of each pepper planting plot; based on the clustering distribution results, the growth stages of each pepper planting plot are divided to obtain the growth status category of each pepper planting plot.

Images