A landscaping detection system

Abstract

The application relates to the technical field of garden detection, and relates to a garden greening detection system. The system comprises a graph generation module, an abnormal area marking module, an electrical conductivity modeling module, a target point positioning module, a sample analysis module and a decision-making module. The system rasterizes target garden area crown layer multi-spectral response data, constructs a vegetation spectral response graph, marks a matrix trait suspicious area, generates a soil electrical conductivity distribution graph, determines high-probability abnormal target points and a hierarchical sampling scheme through spatial overlay analysis, tests the physical and chemical properties of collected hierarchical soil samples, obtains a hierarchical soil test data set, constructs a matrix trait feature vector according to the hierarchical soil test data set, and carries out similarity evaluation on the matrix trait feature vector and a preset healthy soil standard feature vector to obtain a detection analysis conclusion. The application can improve the efficiency of garden greening detection.

Description

Technical Field

[0001] This invention relates to the field of landscape testing technology, and in particular to a landscape greening testing system. Background Technology

[0002] Current methods for monitoring landscaping and greening rely heavily on a combination of manual inspections and fixed-point sampling. This approach not only has limited coverage but also makes it difficult to quickly identify continuous areas with abnormal vegetation spectral responses. Traditional methods lack precise data support for assessing soil matrix properties, and the marking of abnormal areas is highly subjective. They also fail to effectively correlate the spatial relationship between vegetation spectra and soil electrical conductivity, resulting in delayed localization of high-probability anomaly targets and an inability to fully reflect the true condition of landscaping soil.

[0003] Existing technologies lack systematic integration in data processing and analysis. Processes such as rasterization of multispectral data and surface domain construction of electrical conductivity data are fragmented, stratified sampling schemes lack scientific basis, and the utilization rate of soil physicochemical property test results is low. This makes the testing process time-consuming and labor-intensive, and the accuracy and timeliness of the test results are insufficient, failing to meet the needs of refined management of landscaping. Therefore, improving the efficiency of landscaping testing has become an urgent problem to be solved. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides a landscaping detection system, characterized in that the system includes a map generation module, an anomaly area marking module, a conductivity modeling module, a target location module, a sample analysis module, and a decision-making module, wherein: The map generation module is used to rasterize the canopy multispectral response data of the target garden area in order to construct the vegetation spectral response map of the target garden area. The anomaly area marking module is used to identify continuous areas with abnormal spectral response in the vegetation spectral response map and mark the continuous areas as areas of suspicion of matrix properties in the target garden area; The electrical conductivity modeling module is used to construct a surface region of the volume electrical conductivity of the surface soil in areas with questionable matrix properties, and generate a soil electrical conductivity distribution map of the areas with questionable matrix properties. The target location module is used to perform spatial overlay analysis of the gradient change boundary in the vegetation spectral response map and the abnormal high value area in the soil conductivity distribution map to obtain high probability abnormal target points in the target garden area, and to determine the stratified sampling scheme of the target garden area based on the high probability abnormal target points. The sample analysis module is used to perform physicochemical property tests on the collected stratified soil samples according to the stratified sampling scheme, and obtain the stratified soil test dataset of the target garden area; The decision-making module is used to construct matrix trait feature vectors of high-probability abnormal targets based on the stratified soil test dataset, and to evaluate the similarity between the matrix trait feature vectors and the preset healthy soil standard feature vectors to obtain the detection and analysis conclusions of the target garden area.

[0005] In a preferred embodiment, when the map generation module performs rasterization conversion on the canopy multispectral response data of the target garden area to construct the vegetation spectral response map of the target garden area, it is specifically used for: Collect canopy multispectral reflectance data of the target garden area; Geometric correction is performed on the multispectral reflectance data to obtain the corrected multispectral data of the target garden area; Based on the reflectance values ​​of pixels in different spectral bands in the corrected multispectral data, the continuous spectral response curve of the pixel is reconstructed. Spatial gridding is performed on the reflectance values ​​of the same spectral band on the continuous spectral response curve to obtain a single-channel raster layer of the target garden area; By overlaying single-channel raster layers, a vegetation spectral response map of the target garden area is obtained.

[0006] In a preferred embodiment, when the anomaly marking module identifies continuous regions with abnormal spectral responses in the vegetation spectral response map and marks these continuous regions as areas of suspicion regarding the matrix properties of the target garden area, it is specifically used for: Based on the preset healthy vegetation spectral response threshold, principal component analysis is performed on the pixels in the vegetation spectral response map to obtain reference data for the target garden area. Based on the reference data, abnormal response pixels with response values ​​lower than the healthy vegetation spectral response threshold are extracted from the vegetation spectral response map. Neighborhood aggregation is performed on anomalous response pixels to obtain candidate anomalous regions for the target garden area; Logical judgment is made on the area and shape factors of candidate abnormal regions to obtain suspicious areas of matrix properties in the target garden area.

[0007] In a preferred embodiment, when the anomaly marking module performs logical judgment on the area and shape factors of candidate anomaly regions to obtain suspicious areas of matrix properties in the target garden area, it is specifically used for: Obtain the geometric area and shape contour of the candidate anomaly region; The aspect ratio and roundness of the shape profile are parameterized to obtain the shape factor of the shape profile. The geometric area and shape factor are compared with the preset geometric feature thresholds, and candidate regions that meet the geometric feature thresholds are selected from the candidate anomaly regions. The candidate areas are spatially merged to obtain the areas of doubt regarding the matrix properties of the target garden area.

[0008] In a preferred embodiment, when the conductivity modeling module performs surface region construction on the volume conductivity of the surface soil in the suspected matrix property area to generate a soil conductivity distribution map of the suspected matrix property area, it is specifically used for: Soil conductivity measurement points were set up within the spatial range of the area with questionable matrix properties, and the original volume conductivity values ​​of the surface soil at the soil conductivity measurement points were collected. Spatial statistical analysis was performed on the original measurements of bulk conductivity to obtain the spatial autocorrelation structure of the region of doubt regarding matrix properties. Based on the spatial autocorrelation structure, kriging interpolation was performed on the conductivity values ​​between soil conductivity measurement points to obtain a continuous surface conductivity dataset for areas with questionable matrix properties. The continuous surface conductivity dataset was rasterized to obtain a soil conductivity raster layer for areas with questionable matrix properties. Color grading rendering was performed on the soil electrical conductivity raster layer to obtain the soil electrical conductivity distribution map of the area with questionable matrix properties.

[0009] In a preferred embodiment, when the target localization module performs spatial overlay analysis of the gradient change boundary in the vegetation spectral response map and the abnormal high-value region in the soil conductivity distribution map to obtain high-probability abnormal target points in the target garden area, and determines the stratified sampling scheme for the target garden area based on the high-probability abnormal target points, it is specifically used for: Edge detection is performed on the spatial variation gradient of the spectral response values ​​in the vegetation spectral response map to obtain the gradient variation boundary of the vegetation spectral response map; Adaptive threshold segmentation of soil electrical conductivity distribution map is used to obtain abnormally high value areas in the target garden area; By spatially overlaying the gradient change boundary with the abnormal high value region, the initial abnormal intersection area of ​​the target garden area is obtained. Fill the internal holes and small fractures in the initial anomaly intersection area to obtain the target anomaly area of ​​the target garden area; By centroidally locating the target anomaly region, high-probability anomalous target points within the target anomaly region are obtained. Using high-probability abnormal target points as the sampling center points, vertical sampling points are deployed in different depth directions to generate a layered sampling scheme for the target abnormal area.

[0010] In a preferred embodiment, when the target localization module performs adaptive threshold segmentation on the soil conductivity distribution map to obtain abnormally high value areas in the target garden area, it is specifically used for: Histogram statistics were performed on the conductivity values ​​in the soil conductivity distribution map to obtain the frequency distribution of the conductivity values; Based on the frequency distribution, calculate the segmentation threshold used to distinguish normal and abnormal areas in the target garden area; Pixels with conductivity values ​​higher than the segmentation threshold in the soil conductivity distribution map are marked as candidate high-value pixels. Morphological erosion is performed on the candidate high-value pixels to obtain the purified high-value pixels. Spatial connectivity analysis was performed on the high-value pixels after purification to identify the abnormally high-value areas in the target garden area.

[0011] In a preferred embodiment, when the sample analysis module performs physicochemical property tests on the collected stratified soil samples according to the stratified sampling scheme to obtain a stratified soil test dataset of the target garden area, it is specifically used for: Based on the sampling depth marked in the stratified sampling scheme, the collected original soil samples were subjected to depth screening to obtain the soil samples to be tested from the target garden area; Simultaneous determination of soil physicochemical parameters of the soil samples to be tested, to obtain data on the acidity, alkalinity and salinity of the soil samples; The organic matter content and available nitrogen, phosphorus, and potassium content in the soil sample to be tested were determined to obtain the fertility characteristic data of the soil sample. Particle size-density correlation analysis was performed on the soil samples to obtain the physical structure data of the soil samples. The acidity / alkalinity and salinity data, fertility characteristics data, and physical structure data were structured to obtain a layered soil test dataset for the target garden area.

[0012] In a preferred embodiment, when the decision-making module constructs the matrix trait feature vector of high-probability anomalous target points based on the stratified soil test dataset, it is specifically used for: Key parameters characterizing soil chemical properties and physical structure are extracted from the stratified soil test dataset to generate the original parameter set corresponding to high-probability anomaly targets; Eliminate the numerical scale differences between different parameters in the original parameter set to generate a normalized parameter set of the original parameter set; The normalized parameter set is arranged in a structured manner to obtain an ordered parameter sequence of the normalized parameter set; Tensor synthesis is performed on ordered parameter sequences to obtain matrix morphology feature vectors of high-probability anomalous targets.

[0013] In a preferred embodiment, when the decision-making module performs a similarity assessment between the matrix trait feature vector and a preset healthy soil standard feature vector to obtain the detection and analysis conclusion of the target garden area, it is specifically used for: Residual analysis was performed between the matrix trait feature vector and the preset healthy soil standard feature vector to obtain the difference vector of the target garden area. Descriptive statistics are extracted from the difference vectors to obtain their statistical distribution characteristics. Based on statistical distribution characteristics, cluster analysis is performed on the difference vectors to obtain cluster category labels for the difference vectors; Based on the cluster category labels, a level mapping is performed on high-probability anomalous targets to obtain the anomalous level of high-probability anomalous targets. The spatial distribution and anomaly level of high-probability abnormal targets are encapsulated into the detection and analysis conclusions of the target garden area.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention utilizes a map generation module to rasterize and overlay layers on canopy multispectral response data, accurately constructing vegetation spectral response maps and providing a high-quality data foundation for anomaly identification. Through principal component analysis and neighborhood aggregation technology in the anomaly area marking module, it achieves precise screening and spatial merging of suspicious areas related to matrix properties. Using Kriging interpolation and color-grading rendering in the conductivity modeling module, it generates a high-precision soil conductivity distribution map, effectively linking soil and vegetation anomalies. The spatial overlay analysis and centroid positioning technology in the target location module accurately pinpoint high-probability anomaly targets, making the stratified sampling scheme more scientific and targeted, thus improving detection accuracy.

[0015] 2. This invention utilizes a sample analysis module to simultaneously measure the multi-dimensional physicochemical properties of soil, constructing a complete stratified soil test dataset to comprehensively characterize soil matrix properties. The decision-making module, through normalization processing, tensor synthesis, and similarity assessment, achieves precise mapping of anomaly levels, ensuring both objectivity and practicality in the test results. The modules work collaboratively to complete the entire systematic operation from data collection and analysis to conclusion output, significantly improving the accuracy and efficiency of landscaping testing and providing reliable data support and scientific basis for landscaping soil management. Attached Figure Description

[0016] Figure 1 This is a system architecture diagram of a landscaping detection system provided in an embodiment of the present invention.

[0017] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments belong to some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “said” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0020] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0021] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0022] In practice, the server-side equipment deployed in a landscaping and greening inspection system may consist of one or more devices. The aforementioned landscaping and greening inspection system can be implemented as: a business instance, a virtual machine, and hardware devices. For example, this landscaping and greening inspection system can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, this landscaping and greening inspection system can be understood as software deployed on a cloud node, used to provide a landscaping and greening inspection system to various user terminals. Alternatively, this landscaping and greening inspection system can also be implemented as a virtual machine deployed on one or more devices in a cloud node. This virtual machine contains application software for managing various user terminals. Or, this landscaping and greening inspection system can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide a landscaping and greening inspection system to various user terminals.

[0023] In terms of implementation, the landscaping and greening detection system and the user terminal are mutually compatible. That is, if the landscaping and greening detection system is implemented as an application installed on a cloud service platform, then the user terminal is implemented as a client that establishes a communication connection with the application; or if the landscaping and greening detection system is implemented as a website, then the user terminal is implemented as a webpage; or if the landscaping and greening detection system is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0024] like Figure 1 The figure shown is a system architecture diagram of a landscaping detection system provided in an embodiment of the present invention.

[0025] The landscaping detection system 100 described in this invention can be installed on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed on the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed into a website. Depending on the functions implemented, the landscaping detection system 100 may include a map generation module 101, an anomaly area marking module 102, a conductivity modeling module 103, a target location module 104, a sample analysis module 105, and a decision-making module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by an electronic device's processor and perform a fixed function, stored in the electronic device's memory.

[0026] In this embodiment of the invention, in a landscape greening detection system, each of the above-mentioned modules can be implemented independently and can call other modules. Here, "calling" can be understood as one module connecting to multiple modules of another type and providing corresponding services to those connected modules. In the landscape greening detection system provided by this embodiment of the invention, the applicable scope of the landscape greening detection system architecture can be adjusted by adding modules and directly calling them without modifying the program code, achieving cluster-based horizontal expansion to quickly and flexibly expand the landscape greening detection system. In practical applications, the above-mentioned modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0027] The following describes, with reference to specific embodiments, each component of a landscaping and greening detection system and its specific workflow: The map generation module 101 is used to perform rasterization conversion on the canopy multispectral response data of the target garden area in order to construct the vegetation spectral response map of the target garden area. In this embodiment of the invention, when the map generation module performs rasterization conversion on the canopy multispectral response data of the target garden area to construct the vegetation spectral response map of the target garden area, it is specifically used for: Collect canopy multispectral reflectance data of the target garden area; Geometric correction is performed on the multispectral reflectance data to obtain the corrected multispectral data of the target garden area; Based on the reflectance values ​​of pixels in different spectral bands in the corrected multispectral data, the continuous spectral response curve of the pixel is reconstructed. Spatial gridding is performed on the reflectance values ​​of the same spectral band on the continuous spectral response curve to obtain a single-channel raster layer of the target garden area; By overlaying single-channel raster layers, a vegetation spectral response map of the target garden area is obtained.

[0028] A drone equipped with a multispectral sensor was used to collect data on the target garden area. The drone flew along a preset route, which was planned to ensure complete coverage of the boundary of the target garden area. The flight altitude was set to 100 meters, and the sensor maintained a vertical downward shooting angle. Data was collected every 5 meters. During the collection process, the sensor recorded the reflectance values ​​of the vegetation canopy at each shooting position on 30 preset spectral bands. Finally, the canopy multispectral reflectance data covering the entire target garden area was obtained.

[0029] Using a 1:500 high-precision topographic map of the target garden area as a reference, at least 20 clear landmarks are selected from the multispectral reflectance data. These landmarks include feature points with clear spatial locations, such as road corners, corners of fixed structures within the garden, and the base of large tree trunks. The planar coordinates of each landmark on the topographic map and its pixel coordinates in the multispectral reflectance data are read. By adjusting the pixel coordinates of each pixel in the multispectral reflectance data, the pixel coordinates of all selected landmarks are made to perfectly match the planar coordinates on the topographic map, thus completing the geometric correction operation and obtaining the corrected multispectral data of the target garden area.

[0030] Thirty consecutive spectral bands are preset, with a fixed wavelength range for each band. The shortest wavelength is 400 nm, the longest wavelength is 900 nm, and the wavelength interval between adjacent bands is 17 nm. The reflectance value of each pixel in these 30 spectral bands is extracted from the corrected multispectral data. The 30 reflectance values ​​of each pixel are connected sequentially with straight lines according to the wavelength from 400 nm to 900 nm to form a smooth and continuous curve. This curve is the continuous spectral response curve of the pixel.

[0031] The target garden area is divided into uniform grid cells of size 1m x 1m. Each grid cell corresponds to a unique spatial location number. The reflectance values ​​of the same spectral band on the continuous spectral response curve of all pixels are extracted. For example, for the spectral band of 400 nm wavelength, the reflectance values ​​of the corresponding pixels of all grid cells in this band are collected. These reflectance values ​​are filled into the corresponding grid cells respectively. Each spectral band forms a separate raster layer containing the reflectance information of all grid cells. This layer is the single-channel raster layer of the target garden area.

[0032] All 30 single-channel raster layers are superimposed in ascending order of spectral band wavelength. During superposition, the spatial position number of the grid cell is used as the reference to ensure that the position of each grid cell in different single-channel raster layers is completely aligned without any spatial offset. After superposition, each grid cell contains reflectance information of 30 spectral bands. The final comprehensive map containing the spectral information of all vegetation canopies in the target garden area is the vegetation spectral response map of the target garden area.

[0033] The beneficial effects are that, through standardized acquisition methods, precise geometric correction, orderly curve reconstruction, unified standard gridding, and precise alignment of layer overlay, the reflectance information of the vegetation canopy in the target garden area in each spectral band is fully preserved, ensuring the spatial accuracy and spectral integrity of the vegetation spectral response map. This provides high-quality and reproducible basic data for the subsequent anomaly area marking module. The entire process has clear operation steps, and the benchmarks and standards of each link are clear, ensuring the repeatability and stability of the technical solution.

[0034] The abnormal area marking module 102 is used to identify continuous areas with abnormal spectral response in the vegetation spectral response map and mark the continuous areas as suspicious areas of matrix properties in the target garden area; In this embodiment of the invention, when the abnormal area marking module identifies continuous areas with abnormal spectral responses in the vegetation spectral response map and marks these continuous areas as areas of suspicion regarding the matrix properties of the target garden area, it is specifically used for: Based on the preset healthy vegetation spectral response threshold, principal component analysis is performed on the pixels in the vegetation spectral response map to obtain reference data for the target garden area. Based on the reference data, abnormal response pixels with response values ​​lower than the healthy vegetation spectral response threshold are extracted from the vegetation spectral response map. Neighborhood aggregation is performed on anomalous response pixels to obtain candidate anomalous regions for the target garden area; Logical judgment is made on the area and shape factors of candidate abnormal regions to obtain suspicious areas of matrix properties in the target garden area.

[0035] When the anomaly marking module performs logical judgments on the area and shape factors of candidate anomaly regions to obtain suspicious areas of matrix properties in the target garden area, it is specifically used for: Obtain the geometric area and shape contour of the candidate anomaly region; The aspect ratio and roundness of the shape profile are parameterized to obtain the shape factor of the shape profile. The geometric area and shape factor are compared with the preset geometric feature thresholds, and candidate regions that meet the geometric feature thresholds are selected from the candidate anomaly regions. The candidate areas are spatially merged to obtain the areas of doubt regarding the matrix properties of the target garden area.

[0036] The preset healthy vegetation spectral response threshold was determined by collecting and statistically analyzing spectral data from 100 different types of healthy garden vegetation, covering 30 spectral bands. Each band corresponds to a fixed range of reflectance values. The reflectance values ​​of the 30 spectral bands for each pixel in the vegetation spectral response map were arranged into a data matrix by rows and columns, with each pixel as a row and each band reflectance value as a column. By integrating the correlation information of the data from each band, redundant spectral information was removed, and the core data features that could reflect the health status of the vegetation were retained. The resulting simplified dataset is the reference data for the target garden area.

[0037] Each pixel in the reference data corresponds to a set of simplified spectral response feature values. The feature values ​​of each pixel are compared with the preset healthy vegetation spectral response threshold band by band. The feature values ​​of each band must fall within the threshold range of the corresponding band. If the feature value of any band of a pixel does not reach the lower limit of the threshold of that band, the pixel is identified as an abnormal response pixel. The comparison and judgment of all pixels are completed one by one, and all pixels that meet the abnormal conditions are extracted to obtain the abnormal response pixels of the target garden area.

[0038] Each pixel's neighborhood is defined as a 3×3 grid, meaning that the eight pixels surrounding each pixel in the top, bottom, left, right, and diagonal directions are its neighboring pixels. Each abnormal response pixel is checked to see if there are other abnormal response pixels in its neighborhood. If so, these interconnected abnormal response pixels are grouped into a whole to form an independent continuous region. All continuous regions formed in this way are the candidate abnormal regions of the target garden area.

[0039] By counting the total number of pixels contained in each candidate anomaly region and combining the actual spatial area of ​​1 square meter corresponding to each pixel, the geometric area of ​​each candidate anomaly region is calculated by multiplying the total number of pixels by 1 square meter. Along the outermost pixel edge of each candidate anomaly region, starting from any outer pixel, adjacent outer pixels are connected in a clockwise direction until returning to the starting pixel, forming a closed line. This closed line is the shape outline of the candidate anomaly region.

[0040] Select the two points furthest apart on the shape profile and measure the straight-line distance between them as the profile length. Then select the two points furthest apart in the direction perpendicular to the length and measure their straight-line distance as the profile width. Divide the length value by the width value to obtain the aspect ratio of the shape profile. Measure the perimeter of the shape profile and the area it encloses. Multiply the area of ​​the profile by 4π and divide by the square of the perimeter to calculate the roundness of the shape profile. Combine the aspect ratio and roundness into a set of fixed parameters to form a shape factor that can comprehensively characterize the features of the shape profile.

[0041] The preset geometric feature thresholds include an area threshold of 10 square meters, an aspect ratio threshold of 5 in the shape factor, and a roundness threshold of 0.3. The geometric area of ​​each candidate abnormal region is compared with 10 square meters, and the aspect ratio and roundness are compared with 5 and 0.3 in the shape factor. Only candidate abnormal regions with a geometric area greater than or equal to 10 square meters, an aspect ratio less than or equal to 5, and a roundness greater than or equal to 0.3 are selected as candidate regions that meet the geometric feature thresholds.

[0042] The distance threshold for spatial merging is set to 2 meters. The shortest straight-line distance between the edges of the candidate regions after screening is detected one by one. If the shortest straight-line distance between the edges of two candidate regions is less than or equal to 2 meters, the outline edges of the two candidate regions are connected and integrated into a continuous region. All candidate regions are continuously detected until the shortest straight-line distance between the edges of all candidate regions is greater than 2 meters and there are no more regions that can be merged. The continuous region formed in the end is the suspicious area of ​​matrix properties of the target garden area.

[0043] The beneficial effects are that by setting clear thresholds for the spectral response and geometric features of healthy vegetation, combined with specific and operable analysis, calculation and screening steps, abnormal response pixels are accurately extracted and aggregated to form candidate abnormal regions. By using shape parameterization and spatial merging, the accuracy and continuity of suspicious areas of matrix properties are ensured. The entire process is clear, standardized and highly repeatable, effectively avoiding the omission or misjudgment of abnormal areas, and providing accurate and reliable target areas for subsequent soil conductivity detection and target location.

[0044] The electrical conductivity modeling module 103 is used to construct a surface region of the volume electrical conductivity of the surface soil in the area of ​​suspected matrix properties, and generate a soil electrical conductivity distribution map of the area of ​​suspected matrix properties. In this embodiment of the invention, when the conductivity modeling module performs surface region construction on the volume conductivity of the surface soil in the area of ​​suspected matrix properties to generate a soil conductivity distribution map of the area of ​​suspected matrix properties, it is specifically used for: Soil conductivity measurement points were set up within the spatial range of the area with questionable matrix properties, and the original volume conductivity values ​​of the surface soil at the soil conductivity measurement points were collected. Spatial statistical analysis was performed on the original measurements of bulk conductivity to obtain the spatial autocorrelation structure of the region of doubt regarding matrix properties. Based on the spatial autocorrelation structure, kriging interpolation was performed on the conductivity values ​​between soil conductivity measurement points to obtain a continuous surface conductivity dataset for areas with questionable matrix properties. The continuous surface conductivity dataset was rasterized to obtain a soil conductivity raster layer for areas with questionable matrix properties. Color grading rendering was performed on the soil electrical conductivity raster layer to obtain the soil electrical conductivity distribution map of the area with questionable matrix properties.

[0045] Soil conductivity measurement points were set up in the area of ​​suspected matrix properties at a uniform grid spacing of 5 meters × 5 meters. The precise coordinates of each measurement point were recorded by GPS positioning to ensure that all points completely covered the boundary range of the area of ​​suspected matrix properties without any omissions. A portable soil conductivity meter was used to collect the conductivity data of the surface soil at a depth of 0-20 cm at each point. The instrument probe was fully inserted into the soil and kept in close contact with the soil. After the value on the instrument display stabilized for 3 seconds, the value was read and recorded. After all points were measured, the original volumetric conductivity measurement values ​​of the area of ​​suspected matrix properties were collected.

[0046] The original measured values ​​of volume conductivity were correlated with the GPS coordinates of the corresponding measurement points. All measurement points were divided into several 10m × 10m analysis units according to their coordinate positions. The average value of all measurements in each analysis unit was calculated. Then, the difference between the original measured value of each measurement point and the average value of its analysis unit was calculated. The variation of the difference between measurement points within different distance ranges was statistically analyzed. If the difference between points with a distance of less than 3 meters is generally less than the difference between points with a distance of more than 5 meters, it is determined that the conductivity value of the suspicious matrix properties area has a correlation with distance. This correlation is the spatial autocorrelation structure of the suspicious matrix properties area.

[0047] Based on the established spatial autocorrelation structure, using the original measurements of all soil electrical conductivity measurement points as the base data, for the blank areas within the matrix property doubt area where no measurement points were set up, all known measurement points within a 5-meter radius around each blank location were selected. Different weights were assigned to these known points based on their distance from the blank location, with closer points having larger weights. The electrical conductivity value of the known points was multiplied by the corresponding weight and then summed to obtain the estimated electrical conductivity value of the blank location. This method was used to complete the numerical estimation of all blank areas one by one, ultimately forming a continuous surface electrical conductivity dataset covering the entire matrix property doubt area.

[0048] The spatial extent of the area with questionable matrix properties is divided into regular raster cells with a size of 1 meter × 1 meter. Each raster cell corresponds to a unique planar coordinate. The conductivity value corresponding to the center position of each raster cell in the continuous surface conductivity dataset is extracted and used as the attribute value of the raster cell. This ensures that each raster cell has one and only one corresponding conductivity attribute value. All raster cells are arranged and combined in coordinate order to form a soil conductivity raster layer of the area with questionable matrix properties.

[0049] Five fixed electrical conductivity levels are preset: Level 1 is 0-50 μSiemens / cm, Level 2 is 51-100 μSiemens / cm, Level 3 is 101-150 μSiemens / cm, Level 4 is 151-200 μSiemens / cm, and Level 5 is 201 μSiemens / cm and above. Each level is assigned a unique color: blue for Level 1, green for Level 2, yellow for Level 3, orange for Level 4, and red for Level 5. The level is determined based on the electrical conductivity value of each grid cell in the soil electrical conductivity raster layer, and each grid cell is assigned the corresponding color. Finally, a soil electrical conductivity distribution map of the matrix property doubt area is formed, which intuitively presents the electrical conductivity distribution.

[0050] The beneficial effects are that, through clear measurement point layout standards, standardized raw data collection procedures, reasonable spatial statistical analysis methods, accurate interpolation estimation methods, and clear color grading rules, the soil conductivity distribution map can truly, continuously, and intuitively reflect the spatial distribution characteristics of the surface soil conductivity in areas with questionable matrix properties. The data collection and processing process is repeatable and verifiable, providing high-precision and high-reliability basic data support for the identification of abnormally high-value areas in the subsequent target location module.

[0051] The target localization module 104 is used to perform spatial overlay analysis on the gradient change boundary in the vegetation spectral response map and the abnormal high value area in the soil conductivity distribution map to obtain high probability abnormal target points in the target garden area, and determine the stratified sampling scheme of the target garden area based on the high probability abnormal target points. In this embodiment of the invention, when the target localization module performs spatial overlay analysis of the gradient change boundary in the vegetation spectral response map and the abnormal high-value region in the soil conductivity distribution map to obtain high-probability abnormal target points in the target garden area, and determines the stratified sampling scheme for the target garden area based on the high-probability abnormal target points, it is specifically used for: Edge detection is performed on the spatial variation gradient of the spectral response values ​​in the vegetation spectral response map to obtain the gradient variation boundary of the vegetation spectral response map; Adaptive threshold segmentation of soil electrical conductivity distribution map is used to obtain abnormally high value areas in the target garden area; By spatially overlaying the gradient change boundary with the abnormal high value region, the initial abnormal intersection area of ​​the target garden area is obtained. Fill the internal holes and small fractures in the initial anomaly intersection area to obtain the target anomaly area of ​​the target garden area; By centroidally locating the target anomaly region, high-probability anomalous target points within the target anomaly region are obtained. Using high-probability abnormal target points as the sampling center points, vertical sampling points are deployed in different depth directions to generate a layered sampling scheme for the target abnormal area.

[0052] When the target localization module performs adaptive threshold segmentation on the soil conductivity distribution map to obtain abnormally high value areas in the target garden area, it is specifically used for: Histogram statistics were performed on the conductivity values ​​in the soil conductivity distribution map to obtain the frequency distribution of the conductivity values; Based on the frequency distribution, calculate the segmentation threshold used to distinguish normal and abnormal areas in the target garden area; Pixels with conductivity values ​​higher than the segmentation threshold in the soil conductivity distribution map are marked as candidate high-value pixels. Morphological erosion is performed on the candidate high-value pixels to obtain the purified high-value pixels. Spatial connectivity analysis was performed on the high-value pixels after purification to identify the abnormally high-value areas in the target garden area.

[0053] For each pixel in the vegetation spectral response map, a 3×3 neighborhood is defined, which includes the eight neighboring pixels in the top, bottom, left, right, and diagonal directions of the pixel. The difference in spectral response value between the central pixel and each neighboring pixel is calculated one by one. The absolute values ​​of all differences are summed to obtain the spatial gradient value of the pixel. The preset gradient threshold is 0.2. When the spatial gradient value of a pixel is greater than or equal to 0.2, the pixel is marked as an edge pixel. All edge pixels are connected sequentially according to their spatial positions to form a closed or continuous line. This line is the gradient change boundary of the vegetation spectral response map.

[0054] All conductivity values ​​in the soil conductivity distribution map are divided into several continuous intervals at intervals of 10 microsiemens / cm. The interval range starts from 0 microsiemens / cm and increases sequentially to the interval containing the maximum conductivity value in the distribution map. The number of pixels contained in each interval is counted, and the conductivity range and the corresponding number of pixels in each interval are recorded. The statistical results are formed with conductivity interval as the horizontal axis and number of pixels as the vertical axis. This result is the frequency distribution of conductivity values.

[0055] The formula for calculating the segmentation threshold is as follows: ; In the formula, The segmentation threshold is... The arithmetic mean of the conductivity values. These are the preset dispersion weighting coefficients. The standard deviation of the conductivity value. This is the preset skewness compensation coefficient. This represents the total number of pixels in the soil electrical conductivity distribution map. The first one in the soil electrical conductivity distribution map The conductivity value of each pixel.

[0056] Collect the conductivity values ​​of all pixels in the soil conductivity distribution map, add all conductivity values ​​together, and divide by the total number of pixels to obtain the arithmetic mean of the conductivity values. This value is the conductivity value. The total number of pixels is The conductivity value corresponding to each pixel is... Calculate each and The difference is calculated by summing the squares of all the differences and then dividing by . The variance is obtained, and the square root of this variance is taken to obtain the standard deviation of the conductivity value. This value is the standard deviation of the conductivity value. Calculate each and The difference is calculated by summing the cubes of all the differences and then dividing by . Divide by The skewness of the conductivity value is obtained by cubicing the tb. Based on historical datasets of healthy and abnormal soil electrical conductivity from 100 different garden types and soil textures, the accuracy of anomaly area identification was determined by analyzing the impact of the dispersion in each data set on the accuracy of anomaly area identification. The degree of agreement between the identification results and actual anomaly areas under different coefficients was statistically analyzed, and a fixed coefficient of 1.5 was determined to achieve an agreement rate of over 95%. Based on the statistical distribution characteristics of historical soil electrical conductivity data, this formula analyzes the interference pattern of distribution skewness on threshold division, tests the correction effect of different compensation coefficients on skewed data, and determines a fixed coefficient of 0.8 that can control the misjudgment rate caused by skewed data to within 5%. This formula integrates the average level, dispersion, and distribution skew characteristics of electrical conductivity values, considering both the basic conductivity level of most pixels and the impact of data dispersion and distribution asymmetry. This avoids misjudging or missing abnormal areas due to relying on a single indicator to determine the threshold, enabling the obtained segmentation threshold to accurately divide normal conductivity areas from abnormally high value areas. Add 1.5 times Adding 0.8 times the skewness, the final value is the segmentation threshold used to distinguish between normal and abnormal regions.

[0057] The conductivity value of each pixel in the soil conductivity distribution map is read one by one. The value is compared with the calculated segmentation threshold. If the conductivity value of a pixel is greater than the segmentation threshold, the pixel is assigned a unique identifier, which is a high-value candidate. The set of all pixels assigned this identifier is the candidate high-value pixels of the target garden area.

[0058] A 3×3 structuring element is used for morphological erosion. The center of the structuring element is the effective point of action, and the surrounding 8 positions are auxiliary verification points. Each candidate high-value pixel is traversed. Only when the pixel is the center of the structuring element, and at least 6 of the pixels corresponding to the surrounding 8 auxiliary verification points are candidate high-value pixels, is the candidate high-value pixel retained. Otherwise, the pixel is removed. All the retained pixels are the purified high-value pixels of the candidate high-value pixels.

[0059] The criteria for determining spatial connectivity are that there are adjacent relationships between pixels in the vertical, horizontal, or diagonal directions. Starting from the first purified high-value pixel, all purified high-value pixels adjacent to it are searched, these pixels are grouped into a connected group and assigned a unique number. Then, one is selected from the unclassified purified high-value pixels, and the above search and classification operation is repeated until all purified high-value pixels are classified. The connected group corresponding to each number is an independent connected domain. Connected domains containing more than or equal to 20 pixels are selected. These connected domains together constitute the abnormal high-value area of ​​the target garden area.

[0060] Using the spatial coordinates of 1m×1m grid cells as a unified benchmark, the area enclosed by the gradient change boundary is superimposed with the abnormal high value area. Grid cells that belong to both areas are retained, while grid cells that belong to only one area are removed. The continuous area formed by combining all the retained grid cells according to their spatial location is the initial abnormal intersection area of ​​the target garden area.

[0061] The criteria for identifying internal holes are defined as blank grid areas with an area of ​​less than 5 square meters that are completely surrounded by the initial anomaly intersection area. The criteria for identifying minor breaks are breaks with a width of less than 2 grid units and a length of less than 3 grid units. Using the grid unit attributes of the initial anomaly intersection area, all internal holes and minor breaks that meet the criteria are filled to make the area form a complete and continuous shape. The filled area is the target anomaly area of ​​the target garden area.

[0062] The x and y coordinates of all grid cells in the target anomaly area are statistically analyzed. The sum of all x coordinates is divided by the total number of grid cells to obtain the average x coordinate. The sum of all y coordinates is divided by the total number of grid cells to obtain the average y coordinate. The spatial point determined by the average x and y coordinates is the centroid of the target anomaly area. This centroid is the high-probability anomaly target point in the target anomaly area.

[0063] Using a high-probability anomaly target as the sampling center point, three vertical sampling points are evenly distributed within a radius of 5 meters around it, with an angle of 120 degrees between the sampling points. Three sampling depths are set for each sampling point, namely 0-20 cm, 20-40 cm, and 40-60 cm. The GPS coordinates, sampling depth, and sampling order of each sampling point are clearly defined. This information is compiled into a booklet to form a layered sampling scheme for the target anomaly area.

[0064] The beneficial effects are that by using clear gradient detection standards, frequency statistics rules, threshold calculation parameters and formulas determined based on historical data and scientific analysis, morphological processing parameters and spatial analysis benchmarks, the intersection of gradient change boundaries and abnormal high-value areas can be accurately extracted, effectively purifying interference information and improving regional integrity, achieving accurate positioning of high-probability abnormal targets, and the layout of the stratified sampling scheme is scientific and reasonable, ensuring that the collected soil samples are representative, and providing accurate and reliable sampling basis for subsequent soil physicochemical property testing and the generation of detection conclusions.

[0065] The sample analysis module 105 is used to perform physicochemical property tests on the collected stratified soil samples according to the stratified sampling scheme, and obtain the stratified soil test dataset of the target garden area. In this embodiment of the invention, when the sample analysis module performs physicochemical property tests on the collected stratified soil samples according to the stratified sampling scheme to obtain the stratified soil test dataset of the target garden area, it is specifically used for: Based on the sampling depth marked in the stratified sampling scheme, the collected original soil samples were subjected to depth screening to obtain the soil samples to be tested from the target garden area; Simultaneous determination of soil physicochemical parameters of the soil samples to be tested, to obtain data on the acidity, alkalinity and salinity of the soil samples; The organic matter content and available nitrogen, phosphorus, and potassium content in the soil sample to be tested were determined to obtain the fertility characteristic data of the soil sample. Particle size-density correlation analysis was performed on the soil samples to obtain the physical structure data of the soil samples. The acidity / alkalinity and salinity data, fertility characteristics data, and physical structure data were structured to obtain a layered soil test dataset for the target garden area.

[0066] Based on the three sampling depths of 0-20 cm, 20-40 cm, and 40-60 cm specified in the stratified sampling plan, the collected original soil samples were classified and placed according to the corresponding depth. The original soil samples at each depth were sieved using a stainless steel sieve with a 2 mm aperture. During the sieving process, the samples were gently stirred with a glass rod to remove impurities such as stones and plant roots. The soil samples that passed through the sieve after sieving were placed into sealed containers labeled with the sampling point number and sampling depth. Each container contained only a soil sample from a single depth, and finally, the soil samples to be tested in the target garden area were obtained.

[0067] Take 50 grams of each soil sample to be tested and place it in a clean beaker. Add 100 ml of carbon dioxide-free distilled water, stir well with a glass rod, and let it stand for 30 minutes. Stir every 10 minutes during this period to ensure that the soil and water are fully mixed. Insert the electrode of the pH meter into the mixture. After the display value stabilizes for 2 minutes, read and record the value. This value is the acidity or alkalinity data of the soil sample to be tested. At the same time, use a soil salinity analyzer. Insert the instrument probe into the same mixture. After the value stabilizes, read and record the value. This value is the salinity data of the soil sample to be tested. The two types of data are measured and correlated simultaneously.

[0068] Take 10 grams of the soil sample to be tested and place it in a dry hard glass test tube. Add potassium dichromate solution and sulfuric acid solution, shake well, and then place the test tube in an oil bath at 170-180 degrees Celsius for 5 minutes. After cooling to room temperature, titrate with ferrous sulfate solution until the solution color changes from orange-red to blue-green. Calculate the organic matter content based on the volume of ferrous sulfate solution consumed. Determine the available nitrogen content using the Kjeldahl method, obtaining values ​​through digestion, distillation, and titration. Determine the available phosphorus content using the molybdenum-antimony colorimetric method, determining the value by colorimetry after extraction and color development. Determine the available potassium content using flame photometry, introducing the sample extract into a flame photometer to read the value. Integrate the four types of values ​​to obtain the fertility characteristic data of the soil sample to be tested.

[0069] Take 100g of the soil sample to be tested, first remove coarse particles by passing it through a 2mm sieve, then put the remaining sample into a graduated cylinder containing 500ml of distilled water, add dispersant and stir well. Let it stand for 2 minutes, 10 minutes, 2 hours and 24 hours, and measure the density of the suspension at different depths using a hydrometer. Calculate the proportion of particles of different sizes based on the density change, where particles of 2-0.05mm are sand, particles of 0.05-0.002mm are silt, and particles smaller than 0.002mm are clay. At the same time, use the specific gravity bottle method, weigh 5g of the dried soil sample and put it into a specific gravity bottle of known volume, add distilled water to the mark, and calculate the soil particle density by measuring the mass change. Correlate the particle size proportion with the particle density data to obtain the physical structure data of the soil sample to be tested.

[0070] A two-level classification directory is established based on sampling point number and sampling depth. The first-level directory is named after the sampling point number, and the second-level directory is divided according to depth: 0-20 cm, 20-40 cm, and 40-60 cm. A data table is created under each second-level directory, with columns for pH value, salinity, organic matter content, available nitrogen content, available phosphorus content, available potassium content, sand content, silt content, clay content, and particle density. Each type of measurement data is filled into the corresponding cell to ensure that each data is accurately associated with the sampling point and sampling depth. After all the data from all sampling points are processed, a stratified soil test dataset for the target garden area is formed.

[0071] The beneficial effects are that by establishing clear sample processing standards, standardized measurement methods, and precise data analysis procedures, comprehensive multi-dimensional data on soil acidity, alkalinity, salinity, fertility, and physical structure can be obtained. The structured organization method ensures the integrity and relevance of the data, and the operation standards of each step are unified and reproducible. This provides comprehensive, accurate, and reliable data support for the subsequent decision-making module to construct matrix trait feature vectors, ensuring the scientific nature of the similarity assessment results.

[0072] The decision-making module 106 is used to construct matrix trait feature vectors of high-probability abnormal targets based on the stratified soil test dataset, and to evaluate the similarity between the matrix trait feature vectors and the preset healthy soil standard feature vectors in order to obtain the detection and analysis conclusions of the target garden area.

[0073] In this embodiment of the invention, when the decision-making module constructs the matrix trait feature vector of high-probability abnormal target points based on the stratified soil test dataset, it is specifically used for: Key parameters characterizing soil chemical properties and physical structure are extracted from the stratified soil test dataset to generate the original parameter set corresponding to high-probability anomaly targets; Eliminate the numerical scale differences between different parameters in the original parameter set to generate a normalized parameter set of the original parameter set; The normalized parameter set is arranged in a structured manner to obtain an ordered parameter sequence of the normalized parameter set; Tensor synthesis is performed on ordered parameter sequences to obtain matrix morphology feature vectors of high-probability anomalous targets.

[0074] The decision-making module, when performing a similarity assessment between the matrix trait feature vector and the preset healthy soil standard feature vector to obtain the detection and analysis conclusion of the target garden area, is specifically used for: Residual analysis was performed between the matrix trait feature vector and the preset healthy soil standard feature vector to obtain the difference vector of the target garden area. Descriptive statistics are extracted from the difference vectors to obtain their statistical distribution characteristics. Based on statistical distribution characteristics, cluster analysis is performed on the difference vectors to obtain cluster category labels for the difference vectors; Based on the cluster category labels, a level mapping is performed on high-probability anomalous targets to obtain the anomalous level of high-probability anomalous targets. The spatial distribution and anomaly level of high-probability abnormal targets are encapsulated into the detection and analysis conclusions of the target garden area.

[0075] Key parameters characterizing soil chemical properties and physical structure were extracted from the stratified soil test dataset. Chemical property parameters included pH value, salinity, organic matter content, available nitrogen content, available phosphorus content, and available potassium content. Physical structure parameters included sand content, silt content, clay content, and particle density. The parameters were extracted by classifying the sampling points of high-probability anomaly targets and the sampling depths of 0-20 cm, 20-40 cm, and 40-60 cm to ensure that each sampling depth of each target point corresponds to a complete set of parameter data. All extracted parameters were integrated according to the association relationship of "sampling point-sampling depth-parameter type" to form the original parameter set corresponding to the high-probability anomaly targets.

[0076] The numerical ranges of each key parameter are preset, including pH 4-9, salt content 0-200 μSiemens / cm³, organic matter content 0-50 g / kg, available nitrogen 0-200 mg / kg, available phosphorus 0-50 mg / kg, available potassium 0-300 mg / kg, sand content 0-100%, powder content 0-100%, clay content 0-100%, and particle density 2.5-2.7 g / cm³. The minimum-maximum normalization method is used to process each parameter value in the original parameter set. Each parameter value is subtracted from its minimum value, and then divided by the difference between its maximum and minimum values ​​to obtain the normalized value for each parameter. All normalized values ​​are integrated according to the correlation relationships of the original parameter set to generate a normalized parameter set for the original parameter set.

[0077] Following a fixed order of "pH value - salt content - organic matter content - available nitrogen content - available phosphorus content - available potassium content - sand content - powder content - clay content - particle density", the normalized parameters corresponding to each sampling depth of each high-probability anomaly target point are arranged in a structured manner. First, they are sorted from shallow to deep according to the sampling depth. Within the same depth, they are arranged according to the preset parameter order. Each target point forms a set of linearly arranged data containing three sampling depths and ten parameters for each depth. This linearly arranged data is the ordered parameter sequence of the normalized parameter set.

[0078] Using high-probability anomalous targets as the first dimension, sampling depth as the second dimension, and key parameters as the third dimension, tensor synthesis is performed on the ordered parameter sequence. The first dimension contains the unique identifier information of all high-probability anomalous targets, the second dimension corresponds to the three sampling depths of 0-20 cm, 20-40 cm, and 40-60 cm, respectively, and the third dimension corresponds to the normalized values ​​of ten key parameters. The information from the three dimensions is integrated to form a three-dimensional data structure, which is the matrix morphology feature vector of high-probability anomalous targets.

[0079] The preset healthy soil standard feature vector is constructed based on 1,000 sets of healthy soil sample data of different garden types. Its dimensions are completely consistent with the matrix property feature vector. The parameter types and order corresponding to each dimension are the same. The value of each position in the matrix property feature vector is subtracted from the standard value of the corresponding position in the healthy soil standard feature vector to obtain the difference value of each position. All differences are integrated in the order of the original three-dimensional data structure to form the difference vector of the target garden area.

[0080] Descriptive statistics are extracted from the difference vector. The sum of all differences in the difference vector is calculated and divided by the total number of differences to obtain the mean of the difference vector. All differences are arranged in ascending order, and the value in the middle position is taken as the median of the difference vector. The maximum value in the difference vector is subtracted from the minimum value to obtain the range of the difference vector. The difference between each difference and the mean is calculated, the squares of all differences are added together and divided by the total number of differences, and the square root of the result is taken to obtain the standard deviation of the difference vector. The mean, median, range, and standard deviation are integrated to obtain the statistical distribution characteristics of the difference vector.

[0081] The preset number of clusters is 3. Based on the statistical distribution characteristics of the difference vectors, the mean and standard deviation are used as the core classification criteria. Difference vectors with a mean of 0-0.2 and a standard deviation of 0-0.1 are classified into the first class, difference vectors with a mean of 0.2-0.5 and a standard deviation of 0.1-0.3 are classified into the second class, and difference vectors with a mean greater than 0.5 or a standard deviation greater than 0.3 are classified into the third class. Each difference vector is assigned a unique cluster category label: 1 for the first class, 2 for the second class, and 3 for the third class, thus obtaining the cluster category labels for the difference vectors.

[0082] Establish a mapping relationship between cluster category labels and anomaly levels. Cluster category label 1 corresponds to mild anomaly, label 2 corresponds to moderate anomaly, and label 3 corresponds to severe anomaly. Based on the cluster category label corresponding to the high-probability anomaly target, find the corresponding anomaly level. Each high-probability anomaly target obtains a unique anomaly level, thus obtaining the anomaly level of the high-probability anomaly target.

[0083] The GPS coordinates and corresponding anomaly levels of each high-probability anomaly target point are collected. The target garden area is divided into several management units according to its spatial range. Each management unit contains information on all high-probability anomaly target points within that area. The spatial distribution of target points in each management unit is presented in the form of a coordinate list. The anomaly levels are classified as mild, moderate, and severe, and the number of anomalies is counted. The spatial distribution coordinate list and the anomaly level statistics are integrated to form a structured document, which is the detection and analysis conclusion of the target garden area.

[0084] The beneficial effects are that by establishing clear parameter extraction standards, standardizing normalization processes, establishing fixed arrangement and synthesis rules, conducting precise residual analysis, and establishing clear clustering and hierarchical mapping relationships, the scientific nature of the matrix trait feature vector construction and the accuracy of the difference assessment are ensured. The detection and analysis conclusions comprehensively include the spatial location and degree of anomaly information of high-probability anomaly targets. The process is repeatable and the results are verifiable, providing a direct and reliable decision-making basis for the precise improvement and refined management of garden soils.

[0085] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0086] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A landscaping and greening detection system, characterized in that, The system includes a spectrum generation module, an anomaly region marking module, a conductivity modeling module, a target localization module, a sample analysis module, and a decision-making module, wherein: The map generation module is used to rasterize the canopy multispectral response data of the target garden area in order to construct the vegetation spectral response map of the target garden area. The anomaly area marking module is used to identify continuous areas with abnormal spectral response in the vegetation spectral response map and mark the continuous areas as areas of suspicion of matrix properties in the target garden area; The electrical conductivity modeling module is used to construct a surface region of the volume electrical conductivity of the surface soil in areas with questionable matrix properties, and generate a soil electrical conductivity distribution map of the areas with questionable matrix properties. The target location module is used to perform spatial overlay analysis of the gradient change boundary in the vegetation spectral response map and the abnormal high value area in the soil conductivity distribution map to obtain high probability abnormal target points in the target garden area, and to determine the stratified sampling scheme of the target garden area based on the high probability abnormal target points. The sample analysis module is used to perform physicochemical property tests on the collected stratified soil samples according to the stratified sampling scheme, and obtain the stratified soil test dataset of the target garden area; The decision-making module is used to construct matrix trait feature vectors of high-probability abnormal targets based on the stratified soil test dataset, and to evaluate the similarity between the matrix trait feature vectors and the preset healthy soil standard feature vectors to obtain the detection and analysis conclusions of the target garden area.

2. The landscaping detection system as described in claim 1, characterized in that, When the map generation module performs rasterization conversion on the canopy multispectral response data of the target garden area to construct the vegetation spectral response map of the target garden area, it is specifically used for: Collect canopy multispectral reflectance data of the target garden area; Geometric correction is performed on the multispectral reflectance data to obtain the corrected multispectral data of the target garden area; Based on the reflectance values ​​of pixels in different spectral bands in the corrected multispectral data, the continuous spectral response curve of the pixel is reconstructed. Spatial gridding is performed on the reflectance values ​​of the same spectral band on the continuous spectral response curve to obtain a single-channel raster layer of the target garden area; By overlaying single-channel raster layers, a vegetation spectral response map of the target garden area is obtained.

3. The landscaping detection system as described in claim 1, characterized in that, When the anomaly marking module identifies continuous regions with abnormal spectral responses in the vegetation spectral response map and marks these continuous regions as areas of suspicion regarding the matrix properties of the target garden area, it is specifically used for: Based on the preset healthy vegetation spectral response threshold, principal component analysis is performed on the pixels in the vegetation spectral response map to obtain reference data for the target garden area. Based on the reference data, abnormal response pixels with response values ​​lower than the healthy vegetation spectral response threshold are extracted from the vegetation spectral response map. Neighborhood aggregation is performed on anomalous response pixels to obtain candidate anomalous regions for the target garden area; Logical judgment is made on the area and shape factors of candidate abnormal regions to obtain suspicious areas of matrix properties in the target garden area.

4. The landscaping detection system as described in claim 3, characterized in that, When the anomaly marking module performs logical judgments on the area and shape factors of candidate anomaly regions to obtain suspicious areas of matrix properties in the target garden area, it is specifically used for: Obtain the geometric area and shape contour of the candidate anomaly region; The aspect ratio and roundness of the shape profile are parameterized to obtain the shape factor of the shape profile. The geometric area and shape factor are compared with the preset geometric feature thresholds, and candidate regions that meet the geometric feature thresholds are selected from the candidate anomaly regions. The candidate areas are spatially merged to obtain the areas of doubt regarding the matrix properties of the target garden area.

5. A landscaping detection system as described in claim 1, characterized in that, When the conductivity modeling module performs surface region construction on the volume conductivity of the surface soil in areas with questionable matrix properties to generate a soil conductivity distribution map for these areas, it is specifically used for: Soil conductivity measurement points were set up within the spatial range of the area with questionable matrix properties, and the original volume conductivity values ​​of the surface soil at the soil conductivity measurement points were collected. Spatial statistical analysis was performed on the original measurements of bulk conductivity to obtain the spatial autocorrelation structure of the region of doubt regarding matrix properties. Based on the spatial autocorrelation structure, kriging interpolation was performed on the conductivity values ​​between soil conductivity measurement points to obtain a continuous surface conductivity dataset for areas with questionable matrix properties. The continuous surface conductivity dataset was rasterized to obtain a soil conductivity raster layer for areas with questionable matrix properties. Color grading rendering was performed on the soil electrical conductivity raster layer to obtain the soil electrical conductivity distribution map of the area with questionable matrix properties.

6. The landscaping detection system as described in claim 1, characterized in that, The target localization module performs spatial overlay analysis by combining the gradient change boundary in the vegetation spectral response map with the abnormally high value region in the soil conductivity distribution map to obtain high-probability abnormal target points in the target garden area. Then, using these high-probability abnormal target points as the core, it determines the stratified sampling scheme for the target garden area. Specifically, this is done by: Edge detection is performed on the spatial variation gradient of the spectral response values ​​in the vegetation spectral response map to obtain the gradient variation boundary of the vegetation spectral response map; Adaptive threshold segmentation of soil electrical conductivity distribution map is used to obtain abnormally high value areas in the target garden area; By spatially overlaying the gradient change boundary with the abnormal high value region, the initial abnormal intersection area of ​​the target garden area is obtained. Fill the internal holes and small fractures in the initial anomaly intersection area to obtain the target anomaly area of ​​the target garden area; By centroidally locating the target anomaly region, high-probability anomalous target points within the target anomaly region are obtained. Using high-probability abnormal target points as the sampling center points, vertical sampling points are deployed in different depth directions to generate a layered sampling scheme for the target abnormal area.

7. A landscaping detection system as described in claim 6, characterized in that, When the target localization module performs adaptive threshold segmentation on the soil conductivity distribution map to obtain abnormally high value areas in the target garden area, it is specifically used for: Histogram statistics were performed on the conductivity values ​​in the soil conductivity distribution map to obtain the frequency distribution of the conductivity values; Based on the frequency distribution, calculate the segmentation threshold used to distinguish normal and abnormal areas in the target garden area; Pixels with conductivity values ​​higher than the segmentation threshold in the soil conductivity distribution map are marked as candidate high-value pixels. Morphological erosion is performed on the candidate high-value pixels to obtain the purified high-value pixels. Spatial connectivity analysis was performed on the high-value pixels after purification to identify the abnormally high-value areas in the target garden area.

8. The landscaping detection system as described in claim 1, characterized in that, When the sample analysis module performs physicochemical property tests on the collected stratified soil samples according to the stratified sampling scheme to obtain the stratified soil test dataset of the target garden area, it is specifically used for: Based on the sampling depth marked in the stratified sampling scheme, the collected original soil samples were subjected to depth screening to obtain the soil samples to be tested from the target garden area; Simultaneous determination of soil physicochemical parameters of the soil samples to be tested, to obtain data on the acidity, alkalinity and salinity of the soil samples; The organic matter content and available nitrogen, phosphorus, and potassium content in the soil sample to be tested were determined to obtain the fertility characteristic data of the soil sample. Particle size-density correlation analysis was performed on the soil samples to obtain the physical structure data of the soil samples. The acidity / alkalinity and salinity data, fertility characteristics data, and physical structure data were structured to obtain a layered soil test dataset for the target garden area.

9. A landscaping detection system as described in claim 1, characterized in that, When the decision-making module constructs the matrix trait feature vector of high-probability anomalous target points based on the stratified soil test dataset, it is specifically used for: Key parameters characterizing soil chemical properties and physical structure are extracted from the stratified soil test dataset to generate the original parameter set corresponding to high-probability anomaly targets; Eliminate the numerical scale differences between different parameters in the original parameter set to generate a normalized parameter set of the original parameter set; The normalized parameter set is arranged in a structured manner to obtain an ordered parameter sequence of the normalized parameter set; Tensor synthesis is performed on ordered parameter sequences to obtain matrix morphology feature vectors of high-probability anomalous targets.

10. A landscaping detection system as described in claim 1, characterized in that, The decision-making module, when performing a similarity assessment between the matrix trait feature vector and the preset healthy soil standard feature vector to obtain the detection and analysis conclusion of the target garden area, is specifically used for: Residual analysis was performed between the matrix trait feature vector and the preset healthy soil standard feature vector to obtain the difference vector of the target garden area. Descriptive statistics are extracted from the difference vectors to obtain their statistical distribution characteristics. Based on statistical distribution characteristics, cluster analysis is performed on the difference vectors to obtain cluster category labels for the difference vectors; Based on the cluster category labels, a level mapping is performed on high-probability anomalous targets to obtain the anomalous level of high-probability anomalous targets. The spatial distribution and anomaly level of high-probability abnormal targets are encapsulated into the detection and analysis conclusions of the target garden area.

Images