A soil detection sampling system and method
By acquiring multispectral images through a drone platform, soil monitoring layers were divided and optimized, key sampling points and routes were determined, solving the problems of representativeness and accuracy in soil moisture monitoring in forestry environments, and achieving efficient and accurate soil collection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 河北省邢台生态环境监测中心
- Filing Date
- 2026-02-05
- Publication Date
- 2026-06-19
AI Technical Summary
Existing technologies make it difficult to conduct scientific monitoring and sampling of soil moisture content in forestry environments, especially in areas with severe vegetation cover and high spatial heterogeneity, resulting in insufficient representativeness of sampling points and poor inversion accuracy.
Multispectral images were acquired using a drone platform equipped with a multispectral sensor, soil monitoring layers were divided, and key sampling points were determined by optimizing and inverting multi-scale spectral features and soil moisture distribution maps. Sampling paths that avoid obstructions were planned, and sampling equipment was controlled to collect soil samples.
It enables efficient and accurate soil monitoring and sampling in forestry environments with vegetation cover and high spatial heterogeneity, improving the representativeness of sampling points and the accuracy of soil moisture content inversion.
Smart Images
Figure CN122237994A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of monitoring and sampling, and more specifically, to a soil testing and sampling system and method. Background Technology
[0002] Soil testing and sampling is an important technical means in the fields of environmental monitoring and ecological protection, such as forestry. By collecting and analyzing soil samples from the forestry environment, key indicators such as soil fertility, pollutant content, and water content can be assessed, thereby guiding land management in the forestry environment.
[0003] Current soil sampling methods for forestry environments primarily rely on manual single-point soil drilling, laboratory drying measurements, or single-scale remote sensing image inversion. However, these methods fail to consider the spatial distribution patterns of soil moisture content in forestry environments, and cannot cover highly heterogeneous areas such as steep slopes and dense vegetation, resulting in insufficient representativeness of sampling points and difficulty in forming a comprehensive understanding of soil conditions across the entire region. While remote sensing inversion schemes can acquire large-scale multispectral images, they do not scientifically divide monitoring layers based on the heterogeneity of soil moisture content. They only extract spectral information at a fixed spatial scale, which cannot avoid the interference of vegetation shading on the soil moisture content inversion signal, leading to poor inversion accuracy. Furthermore, they are ill-suited to the dual characteristics of high heterogeneity and strong vegetation shading in forestry environments, making it impossible to scientifically determine sampling points that are both representative and resistant to interference. Therefore, how to conduct soil monitoring and sampling based on soil moisture distribution in forestry environments with severe vegetation shading and extremely high spatial heterogeneity has become a challenge for the industry. Summary of the Invention
[0004] This application provides a soil testing and sampling system and method, which can monitor and sample soil based on soil moisture distribution in forestry environments with severe vegetation cover and high spatial heterogeneity.
[0005] In a first aspect, this application provides a soil testing and sampling method for soil testing and sampling in forestry environments. The method involves pre-installing a multispectral sensor on a drone platform and a positioning sensor on the sampling device. The method includes the following steps: The target soil testing area is scanned, and multispectral images of the target soil testing area are acquired by a drone platform equipped with a multispectral sensor; Based on the multispectral image, the target soil detection area is divided into multiple soil monitoring layers, and the soil moisture distribution map of the target soil detection area is determined. Sampling points for each soil monitoring stratum are determined, and then multispectral information of the multispectral image within different spatial scale windows is extracted with all sampling points as the center, so as to obtain the multiscale spectral characteristics of each sampling point under spatial heterogeneity. The soil moisture content of each soil monitoring layer was optimized and inverted using all multi-scale spectral features and the soil moisture distribution map to obtain the key sampling points of each soil monitoring layer under vegetation cover. Based on all key sampling points, a sampling path is determined for the target soil testing area to avoid obstructions. Then, the sampling device equipped with a positioning sensor is controlled to sample the soil in the target soil testing area along the sampling path.
[0006] In some embodiments, dividing the target soil detection area into multiple soil monitoring layers based on the multispectral image and determining the soil moisture distribution map of the target soil detection area specifically includes: Determine the surface reflectance values and multiple topographic factors for multiple bands of the multispectral image; Based on all surface reflectance values and topographic factors, the target soil detection area is divided into multiple soil monitoring layers; The soil moisture distribution map of the target soil detection area is determined based on the reflectance values of a specific band among all surface reflectance values.
[0007] In some embodiments, determining the sampling points for each soil monitoring stratum specifically includes: The number of candidate sampling points for each soil monitoring stratum is determined based on the soil moisture distribution map. By calculating the number of all candidate sampling points, sampling points for each soil monitoring stratum are obtained.
[0008] In some embodiments, extracting multispectral information of the multispectral image within different spatial scale windows centered on all sampling points to obtain the multiscale spectral features of each sampling point under spatial heterogeneity specifically includes: Determine multiple spatial scale windows with different exploration scales; Using all sampling points as the center, multi-scale spectral analysis was performed on the spatial scale windows of different exploration scales of each soil monitoring stratum to obtain the spectral index characteristic values of each sampling point at different scales. Scale assembly is performed on all spectral index feature values to obtain the multi-scale spectral features of each sampling point under spatial heterogeneity.
[0009] In some embodiments, the soil moisture content of each soil monitoring stratum is optimized and inverted using all multi-scale spectral features and the soil moisture distribution map to obtain the key sampling points of each soil monitoring stratum under vegetation cover, specifically including: The soil moisture distribution map and all multi-scale spectral features were correlated to obtain the spectral-moisture content correlation dataset for each soil monitoring layer under vegetation cover. Multi-scale optimization analysis was performed on the spectral-moisture content association datasets of each soil monitoring stratum to obtain multiple inversion accuracies of soil moisture content under different scale windows when vegetation shading occurred for each soil monitoring stratum. Determine the optimal characterization scale for each soil monitoring stratum under vegetation cover based on all inversion accuracies; Based on all optimal characterization scales, key sampling points for each soil monitoring stratum under vegetation cover were selected from candidate sampling points for each soil monitoring stratum.
[0010] In some embodiments, determining the sampling path for the target soil testing area that avoids obstructions based on all key sampling points specifically includes: Obtain static environmental information of the target soil testing area; By taking all key sampling points as targets and combining them with the static environmental information to perform collision avoidance, a sampling path is obtained for the target soil testing area that avoids obstructing obstacles.
[0011] In some embodiments, controlling a sampling device equipped with a positioning sensor to sample soil from the target soil detection area along the sampling path specifically includes: The sampling path is uploaded to the control system of the sampling device, and the sampling device is self-tested, placed at the starting point of the path, and ready. The sampling device is activated and autonomously travels along the sampling path, acquiring location information in real time based on the onboard positioning sensor. When the sampling equipment arrives at each key sampling point, the soil samples at each key sampling point are collected and packaged by an automated drilling device.
[0012] Secondly, this application provides a soil testing and sampling system for soil testing and sampling in forestry environments. The system includes: The acquisition module is used to scan the target soil detection area and acquire multispectral images of the target soil detection area through a drone platform equipped with a multispectral sensor; The processing module is used to divide the target soil detection area into multiple soil monitoring layers based on the multispectral image and determine the soil moisture distribution map of the target soil detection area; The processing module is also used to determine the sampling points of each soil monitoring layer, and then extract the multispectral information of the multispectral image in different spatial scale windows with all sampling points as the center, so as to obtain the multiscale spectral characteristics of each sampling point under spatial heterogeneity. The processing module is also used to optimize and invert the soil moisture content of each soil monitoring layer through all multi-scale spectral features and the soil moisture distribution map, so as to obtain the key sampling points of each soil monitoring layer under vegetation cover. The execution module is used to determine the sampling path of the target soil detection area that avoids obstructions based on all key sampling points, and then control the sampling device equipped with positioning sensors to sample the soil of the target soil detection area along the sampling path.
[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described soil testing and sampling method.
[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described soil testing and sampling method.
[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: The soil testing and sampling system and method provided in this application first scans the target soil testing area by acquiring multispectral images of the target soil testing area using a drone platform equipped with a multispectral sensor. Based on the multispectral images, the target soil testing area is divided into multiple soil monitoring layers, and a soil moisture distribution map of the target soil testing area is determined. Sampling points for each soil monitoring layer are determined, and multispectral information of the multispectral images within different spatial scale windows is extracted using all sampling points as centers to obtain the multiscale spectral characteristics of each sampling point under spatial heterogeneity. The soil moisture content of each soil monitoring layer is optimized and inverted using all multiscale spectral characteristics and the soil moisture distribution map to obtain key sampling points for each soil monitoring layer under vegetation obstruction. Based on all key sampling points, a sampling path for the target soil testing area that avoids obstruction is determined, and then a sampling device equipped with a positioning sensor is controlled to sample the soil in the target soil testing area along the sampling path.
[0016] Therefore, in the soil testing and sampling method of this application, firstly, the target soil testing area is scanned by acquiring multispectral images of the target soil testing area using a drone platform equipped with a multispectral sensor; based on the multispectral images, the target soil testing area is divided into multiple soil monitoring layers, and a soil moisture distribution map of the target soil testing area is determined; sampling points for each soil monitoring layer are determined, and then multispectral information of the multispectral images within different spatial scale windows is extracted using all sampling points as centers, obtaining the multiscale spectral features of each sampling point under spatial heterogeneity, wherein the multiscale spectral features describe the sampling... The high-dimensional spectral characteristics of the area surrounding the sampling point are used to characterize the spectral properties of the sampling point and its surrounding area, capturing spatial heterogeneity from local to global perspectives and reflecting vegetation growth, moisture content, and soil background influences. Secondly, the soil moisture content of each soil monitoring stratum is optimized and inverted using all multi-scale spectral characteristics and the soil moisture distribution map to obtain key sampling points for each soil monitoring stratum under vegetation obstruction. Based on all key sampling points, a sampling path is determined to avoid obstructions in the target soil detection area, and then the sampling device equipped with a positioning sensor is controlled to sample the soil in the target soil detection area along the sampling path. This scheme allows for soil monitoring and sampling in forestry environments with severe vegetation obstruction and extremely high spatial heterogeneity, based on soil moisture distribution. Attached Figure Description
[0017] Figure 1 This is an exemplary flowchart of a soil testing and sampling method according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the determination of multi-scale spectral features according to some embodiments of this application; Figure 3 These are schematic diagrams illustrating collision avoidance according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a soil testing and sampling system according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing a soil testing and sampling method according to some embodiments of this application. Detailed Implementation
[0018] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.
[0019] refer to Figure 1The figure is an exemplary flowchart of a soil testing and sampling method according to some embodiments of this application. This method is used for soil testing and sampling in forestry environments, and a multispectral sensor is pre-mounted on a drone platform, and a positioning sensor is mounted on the sampling device. The soil testing and sampling method mainly includes the following steps: In step 101, the target soil detection area is scanned, and multispectral images of the target soil detection area are acquired by a drone platform equipped with a multispectral sensor.
[0020] In specific implementation, a rotary-wing UAV equipped with a multispectral sensor containing core bands of blue, green, red, red-edge, and near-infrared light is first selected as the scanning device, along with a radiometric calibration diffuse reflector with 95% reflectivity. Simultaneously, the flight parameters of the scanning device are initialized, and the device is then activated to scan the target soil detection area. Next, a vector boundary map of the target soil detection area is imported into the geographic information system software. A hybrid flight path, combining grid-based and terrain-following methods, is planned, and multiple ground control points with known precise projection coordinates are set. Finally, the scan is performed according to the planned flight path to acquire the original multispectral image data. The obtained multispectral image data is then imported into photogrammetric software (such as Pix4DMapper) for radiometric correction, geometric correction, stitching, and cropping (stitching into a global orthophoto image and cropping according to the regional boundaries), ultimately yielding a multispectral orthophoto image containing projection coordinates. Other implementation methods can also be used in other embodiments, which are not limited here.
[0021] In step 102, the target soil detection area is divided into multiple soil monitoring layers based on the multispectral image, and the soil moisture distribution map of the target soil detection area is determined.
[0022] In some embodiments, dividing the target soil detection area into multiple soil monitoring layers based on the multispectral image and determining the soil moisture distribution map of the target soil detection area can be achieved by the following steps: Determine the surface reflectance values and multiple topographic factors for multiple bands of the multispectral image; Based on all surface reflectance values and topographic factors, the target soil detection area is divided into multiple soil monitoring layers; The soil moisture distribution map of the target soil detection area is determined based on the reflectance values of a specific band among all surface reflectance values.
[0023] In specific implementation, determining the surface reflectance values and multiple topographic factors of the multispectral image across multiple bands can be achieved in the following way: The original digital quantization values of the multispectral image are converted into surface reflectance values through radiometric calibration and atmospheric correction, obtaining surface reflectance data for the blue, green, red, red-edge, and near-infrared bands; simultaneously, based on digital elevation model data, elevation, slope, and aspect topographic factors are extracted using topographic analysis algorithms, thereby obtaining the surface reflectance values and multiple topographic factors of the multispectral image across multiple bands. The surface reflectance value of a band refers to the degree of reflection of incident light by the surface within a specific band range, usually expressed as a percentage or dimensionless, used to reflect the spectral characteristics of surface materials (such as vegetation, soil, water bodies, etc.) in different bands; topographic factors refer to various parameters describing topographic features, including elevation, slope, and aspect. Elevation represents the altitude of a point on the Earth's surface; slope represents the degree of inclination of the Earth's surface, usually expressed as an angle or percentage; aspect represents the direction of inclination of the Earth's surface, usually expressed as an azimuth angle; other methods may be used in other embodiments, which are not limited here.
[0024] In specific implementation, dividing the target soil detection area into multiple soil monitoring layers based on all surface reflectance values and topographic factors can be achieved in the following way: combining multi-band reflectance data with topographic factor data to form a multi-dimensional feature space, and performing cluster analysis on the above multi-dimensional feature space. Based on the clustering results (i.e., the similarity of spectral and topographic features), the target soil detection area is divided into several relatively homogeneous regions, each region serving as a soil monitoring layer. Cluster analysis can be performed using an iterative self-organizing data analysis algorithm. Soil monitoring layering refers to dividing the target soil detection area into multiple relatively homogeneous regions based on the soil's spectral and topographic features, with each region serving as a soil monitoring layer, facilitating subsequent extraction of multispectral information from multispectral images within different spatial scale windows. Other methods can also be used in other embodiments, which are not limited here.
[0025] In specific implementation, determining the soil moisture distribution map of the target soil detection area based on the reflectance values of a specific band among all surface reflectance values can be achieved in the following way: Select the reflectance values of the red band and near-infrared band, which are sensitive to soil moisture, from all surface reflectance values; construct an empirical model of the reflectance ratio relationship between the two bands; and then apply the empirical model to the multispectral image of the target soil detection area to generate a preliminary soil moisture distribution map. Other methods can also be used in other embodiments, which are not limited here.
[0026] It should be noted that the soil moisture distribution map in this application refers to the spatial distribution map of soil moisture content within the target soil testing area; it is used to visually display the distribution of soil moisture content within the target soil testing area, which facilitates the subsequent determination of the number of sampling points for each soil monitoring layer.
[0027] In step 103, the sampling points of each soil monitoring layer are determined, and then the multispectral information of the multispectral image in different spatial scale windows is extracted with all the sampling points as the center, so as to obtain the multiscale spectral characteristics of each sampling point under spatial heterogeneity.
[0028] In some embodiments, determining the sampling points for each soil monitoring stratum can be achieved using the following steps: The number of candidate sampling points for each soil monitoring stratum is determined based on the soil moisture distribution map. By calculating the number of all candidate sampling points, sampling points for each soil monitoring stratum are obtained.
[0029] In practice, determining the number of candidate sampling points for each soil monitoring stratum based on the soil moisture distribution map can be achieved in the following way: First, the soil moisture distribution map is analyzed using spatial analysis techniques to calculate the statistical characteristics of soil moisture content within each soil monitoring stratum. These statistical characteristics include the mean, standard deviation, and coefficient of variation. The standard deviation directly characterizes the absolute variation (i.e., spatial variation) of soil moisture content within each soil monitoring stratum, while the coefficient of variation measures the relative variation within the soil monitoring stratum, overcoming the influence of mean differences. Second, using stratified random sampling, the area proportion of each soil monitoring stratum and the spatial variation of its internal soil moisture content are used as known parameters in the sample size calculation formula to calculate the number of candidate sampling points for each soil monitoring stratum. The sample size calculation formula can be: n_h = (N_h * σ_h) / (Σ(N_i * σ_i)) * N_total, where n_h is the number of samples in the h-th soil monitoring stratum, N_h is the area percentage of that stratum, σ_h is the standard deviation of the moisture content of that stratum, and N_total is the preset total sample size, which is set to 200 in this application; wherein, a soil monitoring stratum with a large coefficient of variation indicates that its internal spatial heterogeneity is strong, and more sampling points need to be allocated to fully capture its variability; conversely, the number of sampling points should be reduced to improve efficiency; other methods can also be used in other embodiments, which are not limited here.
[0030] In specific implementation, the sampling points for each soil monitoring stratum are obtained by deploying the sampling points based on the total number of candidate sampling points. This can be achieved in the following way: In the geographic information system software, a spatially constrained random point deployment algorithm is used, and a minimum point spacing constraint is set to deploy the sampling points for each soil monitoring stratum. The minimum point spacing (e.g., 3-5 meters) needs to be set to avoid spatial autocorrelation caused by excessive point clustering and to ensure the spatial independence of the sampling points. In this application, it is set to 4 meters. Finally, the sampling points for each soil monitoring stratum are generated, and the projected coordinates of each sampling point are output. Other methods can also be used in other embodiments, which are not limited here.
[0031] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart for determining multi-scale spectral features in some embodiments of this application. In this embodiment, multi-spectral information of the multi-spectral image within different spatial scale windows is extracted with all sampling points as the center. The multi-scale spectral features of each sampling point under spatial heterogeneity can be obtained by the following steps: First, in step 1031, multiple spatial scale windows with different exploration scales are determined; Secondly, in step 1032, multi-scale spectral analysis is performed on the spatial scale windows of different exploration scales of each soil monitoring layer, with all sampling points as the center, to obtain the spectral index characteristic values of each sampling point at different scales. Finally, in step 1033, scale assembly is performed on all spectral index feature values to obtain the multi-scale spectral features of each sampling point under spatial heterogeneity.
[0032] In practical implementation, determining multiple spatial scale windows for different exploration scales can be achieved as follows: Using pixels of the multispectral image as the unit, an incremental sequence of spatial scale windows is set, where the window size increases in odd-numbered pixel units. For example, if the first spatial scale window sequence is 1*1, the second is 3*3, the third is 5*5, and so on, thus obtaining multiple spatial scale windows for different exploration scales. The window size is in pixels. Combined with the spatial resolution of the multispectral image (cm / pixel), the actual geographical extent can be calculated. Secondly, based on the spatial resolution of the UAV imagery, the actual geographical extent can be calculated. The size calculation of all spatial scale windows is performed to determine the actual geographical range of the target soil detection area corresponding to each spatial scale window. Here, a spatial scale window refers to a pixel window defined to explore the spectral representativeness of a point in different surrounding ranges when the pixel corresponding to the sampling point on the UAV's multispectral image is taken as the center. The spatial scale window sequence is a continuous spatial window sequence covering a range from a single pixel to a macroscopic area, used to automatically optimize the best observation scale from all possible scales to accurately locate the ice and best overcome vegetation shading and spatial heterogeneity. Other methods can also be used in other embodiments, which are not limited here.
[0033] In practice, taking all sampling points as the center, multi-scale spectral analysis is performed on spatial scale windows of different exploration scales for each soil monitoring layer to obtain the spectral index characteristic values of each sampling point at different scales. This can be achieved in the following way: First, the projected coordinates of each sampling point are matched to the multispectral orthophoto image through coordinate transformation, thereby obtaining the pixel center position corresponding to each sampling point. Coordinate transformation matching can be performed using coordinate transformation functions in the geospatial database. Then, a sampling point is selected as the selected sampling point, and the values of all pixels of the selected sampling point within each spatial scale window are calculated in the five original bands: blue, green, red, red edge, and near-infrared. The reflectance values are calculated, and the arithmetic mean of the reflectance values for each original band is calculated. Then, using the average of these bands as input, a number of predefined vegetation indices are calculated. These predefined vegetation indices include at least the Normalized Difference Vegetation Index (NDVI), the Normalized Moisture Index (NMA), and the Soil-Regulated Vegetation Index (SMA). The set of all the obtained vegetation indices and the arithmetic mean of the reflectance values of all original bands at the selected sampling points within each spatial scale window is used as the spectral index feature values of the selected sampling points at different scales. The spectral index features of the remaining sampling points are then determined. Other methods can be used in other embodiments, which are not limited here.
[0034] It should be noted that the spectral index characteristic values in this application are characteristic values describing the spectral response characteristics within a specific range (i.e., a spatial scale window) around a specific sampling point. These include all the aforementioned vegetation indices and the arithmetic mean of the reflectance values of all original bands within each spatial scale window range for the selected sampling point. Among them, the average reflectance of the original bands (i.e., the average reflectance values of the five bands—blue, green, red, red-edge, and near-infrared—within this window) represents the most basic spectral attribute of the area. Vegetation indices (i.e., normalized difference vegetation index, normalized water index, and soil-regulated vegetation index, etc.) characterize the physical and chemical properties of the land surface (such as vegetation abundance, canopy moisture, and soil background influence). There is a strong physiological correlation between vegetation growth status and water content and soil moisture status in the root zone (i.e., vegetation water shortage stress response). Therefore, the spectral characteristics of the tree canopy can serve as an indirect representation of the underlying soil moisture.
[0035] In specific implementation, the scale assembly of all spectral index feature values to obtain the multi-scale spectral features of each sampling point under spatial heterogeneity can be achieved in the following way: select a sampling point as the selected sampling point, and stitch together all the spectral index feature values corresponding to the selected sampling point according to the size order of the spatial scale window from small to large, thereby obtaining the high-dimensional, continuous multi-scale spectral features of the selected sampling point under spatial heterogeneity; continue to determine the multi-scale spectral features of the remaining sampling points; other methods can also be used in other embodiments, which are not limited here.
[0036] It should be noted that the multi-scale spectral features in this application are high-dimensional spectral features describing the area surrounding the sampling point. They are used to characterize the spectral properties of the sampling point and its surrounding area, capture spatial heterogeneity from local to global perspectives, and reflect the growth status of vegetation, moisture content, and soil background influences. This facilitates the indirect characterization of the moisture status of the underlying soil when inferring soil moisture, thereby improving the reliability of the prediction. For any point on the Earth's surface, its measured value (i.e., spectral value) and the physical meaning it represents strongly depend on the spatial size range of the measurement. Therefore, by analyzing the law of spectral response variation with scale, the optimal observation scale that can most effectively overcome vegetation shading and most stably reflect soil moisture information can be found, thus solving the problem of instability of the inversion model in spatially heterogeneous environments caused by using a single fixed scale.
[0037] In step 104, the soil moisture content of each soil monitoring layer is optimized and inverted using all multi-scale spectral features and the soil moisture distribution map to obtain the key sampling points of each soil monitoring layer under vegetation cover.
[0038] In some embodiments, optimizing and retrieving the soil moisture content of each soil monitoring stratum using all multi-scale spectral features and the soil moisture distribution map to obtain the key sampling points of each soil monitoring stratum under vegetation cover can be achieved through the following steps: The soil moisture distribution map and all multi-scale spectral features were correlated to obtain the spectral-moisture content correlation dataset for each soil monitoring layer under vegetation cover. Multi-scale optimization analysis was performed on the spectral-moisture content association datasets of each soil monitoring stratum to obtain multiple inversion accuracies of soil moisture content under different scale windows when vegetation shading occurred for each soil monitoring stratum. Determine the optimal characterization scale for each soil monitoring stratum under vegetation cover based on all inversion accuracies; Based on all optimal characterization scales, key sampling points for each soil monitoring stratum under vegetation cover were selected from candidate sampling points for each soil monitoring stratum.
[0039] In specific implementation, the soil moisture distribution map and all multi-scale spectral features are correlated to obtain the spectral-moisture content correlation dataset for each soil monitoring stratum under vegetation obstruction. This can be achieved as follows: A subset of candidate sampling points from each soil monitoring stratum is selected as a subset sample point according to the stratified random sampling principle; the selected subset sample points are combined with the soil moisture distribution map to estimate the soil moisture content in the field, thereby obtaining the ground-measured true value of soil moisture content at each subset sample point; the projected coordinates of each subset sample point are precisely registered with the multispectral image, and the multi-scale spectral feature vector corresponding to each subset sample point is extracted; the ground-measured true value of soil moisture content at each subset sample point is correlated with the corresponding multi-scale spectral feature vector. Spectral feature vectors are paired and organized according to the soil monitoring strata to which each subset of sample points belongs, thus constructing a spectral-moisture content association dataset for each soil monitoring stratum under vegetation obstruction. The soil moisture content in the field can be estimated using a ground-penetrating radar or time-domain reflectometer sensor mounted on a UAV, performing low-altitude measurements above the selected subset of sample points. The soil moisture content values for each subset of sample points are obtained as the ground-measured true values through the interaction between electromagnetic wave signals and soil moisture. The spectral-moisture content association dataset describes the relationship between multi-scale spectral features and ground-measured moisture content within the spatial scale region where the sampling points are located. Other methods can also be used in other embodiments, which are not limited here.
[0040] In specific implementation, multi-scale optimization analysis is performed on the spectral-moisture content association datasets of each soil monitoring stratum. The multiple inversion accuracies of soil moisture content under different scale windows during vegetation shading for each soil monitoring stratum can be achieved in the following way: Machine learning methods (such as support vector regression and extreme gradient boosting) are used to optimize the optimal representation scale of each soil monitoring stratum under vegetation shading. First, for each soil monitoring stratum, its association dataset is divided into a training set and a validation set, where the ratio in this application is 7 (training set):3 (validation set). Then, for each soil monitoring stratum, the corresponding multi-scale data within any spatial scale window is used. The model is trained using spectral feature variable training data, and its performance is evaluated on the validation set. The coefficient of determination and root mean square error (RMSE) are used as evaluation metrics for inversion accuracy. The final output is a curve showing the spatial scale window versus inversion accuracy for each soil monitoring stratum, yielding multiple inversion accuracies of soil moisture content at different scale windows under vegetation shading conditions. Inversion accuracy refers to the degree of agreement between the model's predicted and actual values during soil moisture content inversion, typically measured by the coefficient of determination and RMSE. The spatial scale window versus inversion accuracy curve describes the mapping relationship between the size of the spatial scale window and the inversion accuracy. Other implementation methods can also be used in other embodiments, which are not limited here.
[0041] It should be noted that under vegetation shading, the spectral characteristics of vegetation will overlap with those of soil, increasing the uncertainty of the inversion and thus reducing its accuracy. Different spatial scale windows have a significant impact on inversion accuracy: smaller windows may capture more detailed local information but are more susceptible to noise interference; larger windows may smooth out the influence of vegetation, improving inversion accuracy, but may also lose local details. Therefore, through multi-scale optimization analysis, the optimal representation scale for different soil monitoring layers under vegetation shading conditions can be found, thereby improving the inversion accuracy of soil moisture content.
[0042] In specific implementation, determining the optimal representation scale for each soil monitoring stratum under vegetation shading based on all inversion accuracies can be achieved as follows: Select one soil monitoring stratum as the selected soil monitoring stratum; from all inversion accuracies of soil moisture content for the selected soil monitoring stratum, select the spatial scale window with the highest coefficient of determination and the lowest root mean square error as the optimal representation scale for the selected soil monitoring stratum; continue to determine the optimal representation scale for the remaining soil monitoring strata. The optimal representation scale refers to the best spatial scale window found through multi-scale optimization analysis for different soil monitoring strata under vegetation shading conditions. This window can maximize the inversion accuracy, i.e., obtain the highest coefficient of determination and the lowest root mean square error. Because the soil characteristics of different strata are different, the impact of vegetation shading on spectral characteristics is also different. A smaller window may capture more local information, but is more susceptible to vegetation noise interference; a larger window can smooth out the vegetation influence and improve the inversion accuracy, but may lose local details. Therefore, the optimal representation scale is the best choice after balancing these factors and can effectively improve the inversion accuracy of soil moisture content. Other methods can also be used in other embodiments, which are not limited here.
[0043] In specific implementation, the selection of key sampling points for each soil monitoring stratum under vegetation cover from candidate sampling points based on all optimal representation scales can be achieved in the following way: Select a soil monitoring stratum as the selected soil monitoring stratum, calculate the contribution of the multi-scale spectral features corresponding to all sampling points in the selected soil monitoring stratum to the model through the high-precision inversion model trained at the optimal representation scale corresponding to the selected soil monitoring stratum, and take the sampling point with the largest contribution as the key sampling point of the selected soil monitoring stratum, and continue to determine the key sampling points of the remaining soil monitoring stratum; other methods can also be used in other embodiments, which are not limited here.
[0044] It should be noted that the key sampling points in this application refer to the sampling points that contribute the most to the inversion results in the high-precision inversion model with the optimal representation scale under vegetation shading conditions. These points can effectively represent the spectral characteristics and moisture content properties of the corresponding soil monitoring strata, while reducing the uncertainty caused by vegetation disturbance, and providing a reliable foundation for subsequent data analysis and model application.
[0045] In step 105, a sampling path for the target soil detection area that avoids obstructions is determined based on all key sampling points, and then the sampling device equipped with a positioning sensor is controlled to sample the soil in the target soil detection area along the sampling path.
[0046] In some embodiments, determining the sampling path for the target soil testing area that avoids obstructions based on all key sampling points can be achieved using the following steps: Obtain static environmental information of the target soil testing area; By taking all key sampling points as targets and combining them with the static environmental information to perform collision avoidance, a sampling path is obtained for the target soil testing area that avoids obstructing obstacles.
[0047] In specific implementation, the static environmental information of the target soil detection area can be obtained in the following way: extract the spatial location and outline of obstacles such as trees, shrubs, and rocks from the acquired multispectral images of the UAV and the simultaneously acquired high-resolution digital surface model, generate a two-dimensional map containing static obstacle information, and use the above map as the static environmental information of the target soil detection area; wherein, the static obstacle information refers to the information of static obstacles (such as trees, shrubs, and rocks) in the target soil detection area when passing through each key sampling point; the spatial location and outline of obstacles such as trees, shrubs, and rocks can be extracted by obstacle recognition algorithms (such as random forest and semantic segmentation algorithms); other methods can also be used in other embodiments, which are not limited here.
[0048] In specific implementation, all key sampling points are taken as targets, and collision avoidance is performed in conjunction with the static environment information. The sampling path of the target soil detection area that avoids occlusion obstacles can be obtained in the following way: the projected coordinates of all key sampling points are used as path points input into the improved A* path planning algorithm, and the static environment information is used as a safety constraint during path planning, thereby outputting the sampling path of the target soil detection area that avoids occlusion obstacles. (Refer to...) Figure 3 The figure described is a schematic diagram of collision avoidance in some embodiments of this application. The dashed line represents the sampling path of the target soil detection area that avoids occlusion obstacles. The improved A* path planning algorithm balances search efficiency and accuracy by introducing dynamic weight coefficients, reduces redundant calculations by using a 5-neighborhood search strategy, defines the device safety radius to achieve collision avoidance, and uses Bézier curves for path smoothing optimization to finally obtain the sampling path. Other methods can also be used in other embodiments, which are not limited here.
[0049] It should be noted that the sampling path in this application refers to a path that can effectively avoid obstructions and ensure efficient and accurate sampling of key sampling points in each soil monitoring layer of the target soil testing area under vegetation obstruction conditions. It is used to plan and restrict the order of sampling points and facilitate subsequent control of the sampling equipment to sample along the sampling path.
[0050] In some embodiments, controlling a sampling device equipped with a positioning sensor to sample soil from the target soil detection area along the sampling path can be achieved by the following steps: The sampling path is uploaded to the control system of the sampling device, and the sampling device is self-tested, placed at the starting point of the path, and ready. The sampling device is activated and autonomously travels along the sampling path, acquiring location information in real time based on the onboard positioning sensor. When the sampling equipment arrives at each key sampling point, the soil samples at each key sampling point are collected and packaged by an automated drilling device.
[0051] It should be noted that during the sampling process, the monitoring area equipment needs to monitor its own status and the surrounding environment in real time, automatically record the location coordinates, sampling time and other information of each sampling point, and autonomously return to the starting point or designated assembly point after completing the sampling task of all points.
[0052] In another aspect, in some embodiments, this application provides a soil testing and sampling system for soil testing and sampling in forestry environments, with reference to... Figure 4 The figure is a schematic diagram of the structure of a soil testing and sampling system according to some embodiments of this application. The soil testing and sampling system includes: an acquisition module 401, a processing module 402, and an execution module 403, which are described below: The acquisition module 401 in this application is mainly used to scan the target soil detection area and acquire multispectral images of the target soil detection area through a drone platform equipped with a multispectral sensor. Processing module 402, in this application, is used to divide the target soil detection area into multiple soil monitoring layers based on the multispectral image and determine the soil moisture distribution map of the target soil detection area; It should be noted that the processing module 402 in this application is also used to determine the sampling points of each soil monitoring layer, and then extract the multispectral information of the multispectral image in different spatial scale windows with all sampling points as the center, so as to obtain the multiscale spectral characteristics of each sampling point under spatial heterogeneity. In addition, it should be noted that the processing module 402 in this application is also used to optimize and invert the soil moisture content of each soil monitoring layer through all multi-scale spectral features and the soil moisture distribution map, so as to obtain the key sampling points of each soil monitoring layer under vegetation cover. The execution module 403 in this application is mainly used to determine the sampling path of the target soil detection area that avoids obstructions based on all key sampling points, and then control the sampling device equipped with positioning sensors to sample the soil of the target soil detection area along the sampling path.
[0053] In addition, this application also provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described soil testing and sampling method.
[0054] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device for implementing a soil testing and sampling method according to some embodiments of this application. The soil testing and sampling method in the above embodiments can be implemented through... Figure 5 The computer device shown is used to implement this, and the computer device includes at least one processor 501, a communication bus 502, a memory 503, and at least one communication interface 504.
[0055] Processor 501 can be a general-purpose central processing unit (CPU) or an application-specific integrated circuit (ASIC).
[0056] The communication bus 502 can be used to transmit information between the aforementioned components.
[0057] Memory 503 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CDROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. Memory 503 may exist independently and be connected to processor 501 via communication bus 502. Memory 503 may also be integrated with processor 501.
[0058] The memory 503 stores program code for executing the scheme of this application, and its execution is controlled by the processor 501. The processor 501 executes the program code stored in the memory 503. The program code may include one or more software modules. The method used in the above embodiments can be implemented by the processor 501 and one or more software modules in the program code in the memory 503.
[0059] Communication interface 504 uses any transceiver-like device to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.
[0060] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single CPU) processor or a multi-core (multi CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).
[0061] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.
[0062] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described soil testing and sampling method.
[0063] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0064] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A soil testing and sampling method for soil testing and sampling in forestry environments, comprising pre-mounting a multispectral sensor on an unmanned aerial vehicle (UAV) platform and a positioning sensor on the sampling device, characterized in that, The method includes the following steps: The target soil testing area is scanned, and multispectral images of the target soil testing area are acquired by a drone platform equipped with a multispectral sensor; Based on the multispectral image, the target soil detection area is divided into multiple soil monitoring layers, and the soil moisture distribution map of the target soil detection area is determined. Sampling points for each soil monitoring stratum are determined, and then multispectral information of the multispectral image within different spatial scale windows is extracted with all sampling points as the center, so as to obtain the multiscale spectral characteristics of each sampling point under spatial heterogeneity. The soil moisture content of each soil monitoring layer was optimized and inverted using all multi-scale spectral features and the soil moisture distribution map to obtain the key sampling points of each soil monitoring layer under vegetation cover. Based on all key sampling points, a sampling path is determined for the target soil testing area to avoid obstructions. Then, the sampling device equipped with a positioning sensor is controlled to sample the soil in the target soil testing area along the sampling path.
2. The method as described in claim 1, characterized in that, Based on the multispectral imagery, the target soil detection area is divided into multiple soil monitoring layers, and the soil moisture distribution map of the target soil detection area is determined, specifically including: Determine the surface reflectance values and multiple topographic factors for multiple bands of the multispectral image; Based on all surface reflectance values and topographic factors, the target soil detection area is divided into multiple soil monitoring layers; The soil moisture distribution map of the target soil detection area is determined based on the reflectance values of a specific band among all surface reflectance values.
3. The method as described in claim 1, characterized in that, The specific sampling points for each soil monitoring stratum include: The number of candidate sampling points for each soil monitoring stratum is determined based on the soil moisture distribution map. By calculating the number of all candidate sampling points, sampling points for each soil monitoring stratum are obtained.
4. The method as described in claim 1, characterized in that, Extracting multispectral information of the multispectral image within different spatial scale windows using all sampling points as centers, and obtaining the multiscale spectral features of each sampling point under spatial heterogeneity, specifically includes: Determine multiple spatial scale windows with different exploration scales; Using all sampling points as the center, multi-scale spectral analysis was performed on the spatial scale windows of different exploration scales of each soil monitoring stratum to obtain the spectral index characteristic values of each sampling point at different scales. Scale assembly is performed on all spectral index feature values to obtain the multi-scale spectral features of each sampling point under spatial heterogeneity.
5. The method as described in claim 1, characterized in that, The soil moisture content of each soil monitoring stratum was optimized and inverted using all multi-scale spectral features and the soil moisture distribution map. The key sampling points for each soil monitoring stratum under vegetation cover were obtained, including: The soil moisture distribution map and all multi-scale spectral features were correlated to obtain the spectral-moisture content correlation dataset for each soil monitoring layer under vegetation cover. Multi-scale optimization analysis was performed on the spectral-moisture content association datasets of each soil monitoring stratum to obtain multiple inversion accuracies of soil moisture content under different scale windows when vegetation shading occurred for each soil monitoring stratum. Determine the optimal characterization scale for each soil monitoring stratum under vegetation cover based on all inversion accuracies; Based on all optimal characterization scales, key sampling points for each soil monitoring stratum under vegetation cover were selected from candidate sampling points for each soil monitoring stratum.
6. The method as described in claim 1, characterized in that, The sampling path for determining the target soil testing area, which avoids obstructions, based on all key sampling points specifically includes: Obtain static environmental information of the target soil testing area; By taking all key sampling points as targets and combining them with the static environmental information to perform collision avoidance, a sampling path is obtained for the target soil testing area that avoids obstructing obstacles.
7. The method as described in claim 1, characterized in that, Controlling a sampling device equipped with a positioning sensor to sample soil from the target soil detection area along the sampling path specifically includes: The sampling path is uploaded to the control system of the sampling device, and the sampling device is self-tested, placed at the starting point of the path, and ready. The sampling device is activated and autonomously travels along the sampling path, acquiring location information in real time based on the onboard positioning sensor. When the sampling equipment arrives at each key sampling point, the soil samples at each key sampling point are collected and packaged by an automated drilling device.
8. A soil testing and sampling system for soil testing and sampling in forestry environments, characterized in that, The system includes: The acquisition module is used to scan the target soil detection area and acquire multispectral images of the target soil detection area through a drone platform equipped with a multispectral sensor; The processing module is used to divide the target soil detection area into multiple soil monitoring layers based on the multispectral image and determine the soil moisture distribution map of the target soil detection area; The processing module is also used to determine the sampling points of each soil monitoring layer, and then extract the multispectral information of the multispectral image in different spatial scale windows with all sampling points as the center, so as to obtain the multiscale spectral characteristics of each sampling point under spatial heterogeneity. The processing module is also used to optimize and invert the soil moisture content of each soil monitoring layer through all multi-scale spectral features and the soil moisture distribution map, so as to obtain the key sampling points of each soil monitoring layer under vegetation cover. The execution module is used to determine the sampling path of the target soil detection area that avoids obstructions based on all key sampling points, and then control the sampling device equipped with positioning sensors to sample the soil of the target soil detection area along the sampling path.
9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the soil testing and sampling method as described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the soil testing and sampling method as described in any one of claims 1 to 7.