Sampling position determination method and device, equipment and storage medium

Abstract

The application provides a sampling position determination method, device, equipment and storage medium, the method comprises the following steps: obtaining remote sensing data collected by remote sensing quantitative equipment to be verified, and prior data with higher spatial resolution than the remote sensing data; performing hierarchical processing on the prior data and constructing a screening grid, performing multi-level screening on the multi-layer data obtained by hierarchical processing based on the screening grid, and obtaining preselected sampling units in each layer; obtaining the total number of samples and distributing them to each layer of data, determining target sampling units according to the number of samples allocated to each layer of data and the preselected sampling units; and determining the sampling position of the remote sensing data according to the spatial position of the target sampling units. The sampling position determination method provided by the application ensures the spatial homogeneity of the sampling units through hierarchical processing of the prior data and multi-level screening based on the grid, thereby improving the reliability of the authenticity verification result based on the sampling units.

Description

Technical Field

[0001] This invention relates to the field of remote sensing technology, and in particular to a method, apparatus, device, and storage medium for determining sampling locations. Background Technology

[0002] Remote sensing technology, with its wide observation range, continuity, and real-time capabilities, has become an important tool for observing the Earth's environment. Multi-source, multi-scale quantitative remote sensing products are playing an increasingly important role in global climate change and ecological environment research. Verification of authenticity is a crucial step in ensuring the quality of quantitative remote sensing products and supporting their application. Authenticity refers to the ability of quantitative remote sensing products to accurately and truthfully reflect the actual conditions of the Earth's surface. Currently, the most common method for verifying the authenticity of quantitative remote sensing products is based on prior knowledge and involves manual decision-making. This involves setting up basic sampling units on a large, continuously distributed, homogeneous surface, or determining sampling units through on-site sampling before ground experiments. Then, remote sensing data is sampled and used for verification to check the authenticity of the quantitative remote sensing product. However, these verification methods lack quantitative evaluation of the surface homogeneity of the selected sampling units, the sample representativeness, and the spatial representativeness of the sampled data, leading to low reliability of the verification results. Summary of the Invention

[0003] This invention provides a method, apparatus, device, and storage medium for determining sampling locations, in order to solve the technical problem that the sampling units selected in the prior art for remote sensing data do not quantitatively evaluate the homogeneity of the land surface, resulting in a lack of sample representativeness and spatial representativeness in the sampling data, thus leading to low reliability of the authenticity test results.

[0004] This invention provides a method for determining a sampling location, comprising:

[0005] The remote sensing data collected by the remote sensing quantitative device to be verified is obtained, as well as the prior data corresponding to the remote sensing data, wherein the spatial resolution of the prior data is higher than that of the remote sensing data.

[0006] The prior data is processed into layers to obtain multi-layered data corresponding to the prior data;

[0007] Construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data.

[0008] Obtain the total number of samples and allocate the total number of samples to the multi-layered data. Determine the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling unit.

[0009] The sampling location of the remote sensing data is determined based on the spatial location of the target sampling unit.

[0010] According to a sampling location determination method provided by the present invention, the step of performing multi-level filtering on the multi-layered data based on the filtering grid to determine the pre-selected sampling units in each layer of the multi-layered data includes:

[0011] Based on the multi-layered data, calculate the first screening parameter in each grid of the screening grid;

[0012] Obtain a first parameter range, and based on the first filtering parameter and the first parameter range, filter out a first set of grids from each grid of the filtering grid;

[0013] Based on the multi-layered data, calculate the second filtering parameters within a preset grid range, centered on each grid in the first grid set;

[0014] Obtain the second parameter range, and based on the second filtering parameter and the second parameter range, filter out the second grid set from the first grid set;

[0015] The pre-selected sampling units in each layer of the multi-layered data are determined based on each grid in the second grid set, wherein the first filtering parameter and the second filtering parameter include the coefficient of variation and the cell coverage ratio, and each grid of the filtering grid is a basic sampling unit.

[0016] According to a sampling location determination method provided by the present invention, the step of determining a target sampling unit based on the sampling quantity allocated to each layer of the multi-layered data and the pre-selected sampling units includes:

[0017] The number of samples allocated to each layer of the multi-layered data is spatially balanced to obtain the set of sampling units in each layer of the multi-layered data;

[0018] Calculate the Euclidean distance between each pair of sampling units in the sampling unit set, and filter out sampling unit pairs whose Euclidean distance is less than a preset distance threshold;

[0019] Based on the coefficient of variation in the first screening parameters, the sampling unit is determined to be a first sampling unit with a smaller coefficient of variation and a second sampling unit with a larger coefficient of variation.

[0020] Centered on the first sampling unit, search for alternative sampling units for the second sampling unit based on the pre-selected sampling units;

[0021] The second sampling unit is replaced by the alternative sampling unit to obtain the target sampling unit.

[0022] According to a sampling location determination method provided by the present invention, the step of searching for alternative sampling units of the second sampling unit based on the pre-selected sampling unit, with the first sampling unit as the center, includes:

[0023] Obtain search parameters, including search distance and search step size;

[0024] Centered on the first sampling unit, the pre-selected sampling units are searched outwards from the search distance according to the search step size;

[0025] When the pre-selected sampling unit is found, the pre-selected sampling unit with the smallest coefficient of variation is selected from the pre-selected sampling units found, based on the coefficient of variation in the first screening parameter, as the replacement sampling unit for the second sampling unit.

[0026] According to a sampling location determination method provided by the present invention, the step of determining the sampling location of the remote sensing data based on the spatial location of the target sampling unit includes:

[0027] Spatial joint processing is performed on the target sampling unit to determine the spatial position of the target sampling unit in the prior data;

[0028] The spatial location is mapped onto the remote sensing data to determine the sampling location of the remote sensing data.

[0029] According to a sampling location determination method provided by the present invention, the step of performing stratification processing on the prior data to obtain multi-stratified data corresponding to the prior data includes:

[0030] The gradient values ​​of the target indicators to be tested by the remote sensing quantitative device are obtained, and the target indicators are divided into multiple levels based on the gradient values;

[0031] Calculate the target index of the prior data, and perform stratification processing on the prior data according to the level corresponding to the target index of the prior data to obtain multi-layered data corresponding to the prior data.

[0032] According to a sampling location determination method provided by the present invention, before the step of performing hierarchical processing on the prior numbers, the method further includes:

[0033] Based on the target indicators to be tested by the remote sensing quantitative equipment, the areas to be tested and the areas not to be tested in the prior data are determined.

[0034] The non-inspection area is masked, and the area to be inspected is buffered on one side according to the mask boundary.

[0035] The present invention also provides a sampling location determination device, comprising:

[0036] The data acquisition module is used to acquire remote sensing data collected by the remote sensing quantitative device to be verified, as well as prior data corresponding to the remote sensing data, wherein the spatial resolution of the prior data is higher than that of the remote sensing data.

[0037] A hierarchical processing module is used to perform hierarchical processing on the prior data to obtain multi-layered data corresponding to the prior data.

[0038] The grid filtering module is used to construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data.

[0039] The filtering and optimization module is used to obtain the total number of samples and allocate the total number of samples to the multi-layered data. Based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling units, the target sampling unit is determined.

[0040] The location determination module is used to determine the sampling location of the remote sensing data based on the spatial location of the target sampling unit.

[0041] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the sampling position determination method as described above.

[0042] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the sampling position determination method as described above.

[0043] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the sampling position determination method as described above.

[0044] The sampling location determination method, apparatus, device, and storage medium provided by this invention perform stratified processing of prior data, and based on the stratified processing results, conducts multi-level grid-based screening to determine pre-selected sampling units in each stratum. Based on these pre-selected sampling units, the spatial layout of the sampling units allocated in each stratum is optimized to determine the final target sampling units for sampling. Sample data obtained by sampling based on the spatial location of these target sampling units can comprehensively consider the heterogeneity of the sample space and the spatial scale of the remote sensing quantitative equipment under test, ensuring the spatial homogeneity of the sampling units and thus improving the reliability of the authenticity verification results based on these sampling units. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0046] Figure 1 This is one of the flowcharts illustrating the sampling location determination method provided by the present invention;

[0047] Figure 2 This is the second flowchart illustrating the sampling location determination method provided by the present invention;

[0048] Figure 3 This is a schematic diagram of the sampling position determination device provided by the present invention;

[0049] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0051] Reference Figure 1 , Figure 1 This is one of the flowcharts illustrating the sampling location determination method provided by the present invention, based on Figure 1 The sampling location determination method provided by this invention is applied to a remote sensing quantitative device whose authenticity needs to be verified, and includes the following steps:

[0052] Step 100: Obtain remote sensing data collected by the remote sensing quantitative device to be verified, and prior data corresponding to the remote sensing data, wherein the spatial resolution of the prior data is higher than that of the remote sensing data;

[0053] For verifying the authenticity of quantitative remote sensing equipment, a common method is to compare and analyze the equipment's data with reference data that represents the relative true value of ground targets, such as ground-based measured data obtained through field surveys, or higher-resolution remote sensing data acquired by quantitative remote sensing equipment with higher spatial resolution. This comparison evaluates the equipment's ability to accurately and truthfully reflect the actual surface conditions. For medium- to high-resolution equipment, authenticity verification based on ground experiments first requires selecting a homogeneous region with a pixel scale comparable to the equipment under test as the basic sampling unit. Then, the measurement points within the sampling unit are spatially aggregated to increase the scale for authenticity verification. Under the premise of strictly adhering to measurement procedures, the spatially homogeneous basic sampling unit and its reasonable spatial layout are key factors determining the rationality and reliability of the verification results.

[0054] Based on this, a sampling location determination method of the present invention is proposed. This method comprehensively considers the heterogeneity of the sample space and the spatial scale of the remote sensing quantitative device to be tested, ensuring the spatial homogeneity of the sampling unit and thus improving the reliability of the authenticity test results based on the sampling unit. First, remote sensing data collected by the remote sensing quantitative device to be tested, and corresponding prior data, are acquired. The acquired prior data is data from the same phenological period as the remote sensing data, reflects the spatiotemporal characteristics of the parameter to be tested, and has a higher resolution than the remote sensing data from the remote sensing quantitative device. The spatial resolution of the prior data is higher than that of the remote sensing quantitative device to be tested. For example, if the spatial resolution of the remote sensing quantitative device to be tested is 10m, the spatial resolution of the prior data can be 4m. By selecting remote sensing data that is from the same phenological period, reflects the spatiotemporal characteristics of the parameter to be tested, and has a higher spatial resolution than the product to be tested, as prior knowledge, the authenticity of the remote sensing quantitative device to be tested can be determined through comparative analysis.

[0055] Furthermore, the acquired prior data is not only related to the remote sensing quantitative equipment whose authenticity is to be verified, but also to the parameters of the remote sensing quantitative equipment to be verified. For example, the verification of vegetation parameters can select the Normalized Difference Vegetation Index (NDVI) data as prior data, and the verification of spectral parameters can select bands with an approximate spectral range as prior data, which will not be elaborated here.

[0056] Step 200: Perform stratification on the prior data to obtain multi-stratified data corresponding to the prior data;

[0057] The acquired prior data is stratified, divided into different levels according to the parameter to be tested, resulting in multi-layered data. In the visualization of remote sensing data, different layers display different information. For example, using color as the display information, the parameter to be tested is represented by the same color in the same level of remote sensing data, and by different colors in different levels. It should be noted that in multi-layered data, data within the same layer are not necessarily concentrated in the corresponding spatial area of ​​the remote sensing data; they may be dispersed.

[0058] Furthermore, step 200 specifically includes:

[0059] Step 201: Obtain the gradient value of the target indicator to be tested by the remote sensing quantitative device, and divide the target indicator into multiple levels based on the gradient value;

[0060] Step 202: Calculate the target index of the prior data, and perform stratification processing on the prior data according to the level corresponding to the target index of the prior data to obtain multi-layered data corresponding to the prior data.

[0061] When performing stratified processing on prior data, the gradient value of the target indicator to be tested by the remote sensing quantitative equipment is first obtained. Based on the gradient value, the target indicator is divided into multiple levels. The target indicator of the prior data is calculated. The prior data is then stratified according to the level corresponding to the target indicator of the prior data, and the prior data is divided into multiple layers to obtain the corresponding multi-layered data. Among them, the data with the same level corresponding to the target indicator are in the same layer.

[0062] Taking NDVI data as an example, if the gradient value of NDVI is 0.1, then the prior data corresponding to NDVI is divided into multiple levels according to this gradient value, namely Class 0: NDVI(0,0.3), Class 1: NDVI[0.3,0.4), Class 2: NDVI[0.4,0.5), Class 3: NDVI[0.5,0.6), Class 4: NDVI[0.6,0.7), Class 5: NDVI[0.7,0.8), Class 6: NDVI[0.8,0.9), and Class 7: NDVI[0.9,1.0). The NDVI value of the prior data is calculated, and the corresponding NDVI level is determined based on the calculated NDVI value. The prior data is then stratified, that is, prior data of the same NDVI level are grouped into the same layer, thus obtaining multi-layered data.

[0063] Step 300: Construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data;

[0064] A multi-layered data filtering grid is constructed based on the spatial resolution of the remote sensing quantitative instrument under test. For example, the spatial resolution of the constructed grid is three times that of the remote sensing quantitative instrument under test. This constructed filtering grid serves as the basis for selecting sampling units, with each grid cell constituting a basic sampling unit. Based on the constructed filtering grid, multi-level filtering is performed on the multi-layered data obtained from the stratification process to determine the pre-selected sampling units in each layer of the multi-layered data. According to the pre-established grid-based multi-level filtering rules, spatial statistical filtering is performed on each layer of the stratification results to determine the pre-selected basic sampling units and their spatial locations in each layer of the multi-layered data. The filtering conditions for the multi-level filtering include the spatial heterogeneity of the prior data and the spatial scale of the remote sensing quantitative instrument under test.

[0065] Step 400: Obtain the total number of samples and allocate the total number of samples to the multi-layered data. Determine the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling unit.

[0066] The total number of samples is obtained and allocated to each layer of the multi-layered data. Based on the number of samples allocated to each layer and the pre-selected sampling units, the target sampling units are determined. The allocation of the total number of samples to each layer of the multi-layered data is based on a preset principle, including the Neiman allocation principle. The allocation method corresponding to the Neiman allocation principle is shown in Formula 1 below:

[0067]

[0068] Where n represents the total number of samples, which is a configurable parameter and can be preset; L represents the number of strata in the multi-stratified data; n h N represents the number of samples allocated to the h-th layer in a multi-layered dataset of layer L. h σ represents the total number of pixels in the h-th layer. h This represents the standard deviation of the data in the h-th layer.

[0069] Based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling units, the target sampling units are determined. Specifically, based on the selected pre-selected sampling units, the sampling units allocated to each layer are optimized, thereby optimizing the spatial layout of the sampling positions in each layer and obtaining the final sampling units.

[0070] Step 500: Determine the sampling location of the remote sensing data based on the spatial location of the target sampling unit.

[0071] Based on the spatial location of the target sampling units selected through multi-level screening, the sampling location of the remote sensing data is determined. It can be seen that the acquired prior data and the remote sensing data collected by the remote sensing quantitative equipment to be tested have the same spatial range, are close in time, and have the same phenological period. By performing multi-level screening on the prior data, the sampling location of the remote sensing data can be determined based on the spatial location of the selected target sampling units. Site selection and sampling are carried out based on this sampling location. The obtained sample data comprehensively considers multiple factors such as spatial heterogeneity and the spatial scale of the remote sensing quantitative equipment to be tested, making the sample data representative and improving the reliability of the authenticity test results based on the sample data.

[0072] Furthermore, step 500 also includes:

[0073] Step 501: Perform spatial joint processing on the target sampling unit to determine the spatial position of the target sampling unit in the prior data;

[0074] Step 502: Map the spatial location to the remote sensing data to determine the sampling location of the remote sensing data.

[0075] When determining the sampling location of remote sensing data based on the spatial location of the target sampling unit, the specific process involves spatial joint processing of the target sampling units, merging each target sampling unit, determining its spatial location within the spatial range corresponding to the prior data, and mapping based on this spatial location to map each target sampling unit into the remote sensing data, thereby obtaining the sampling location of the remote sensing data.

[0076] Furthermore, prior to step 200, the following steps are also included:

[0077] Step 210: Determine the area to be tested and the non-tested area of ​​the prior data based on the target index to be tested by the remote sensing quantitative equipment;

[0078] Step 220: Mask the non-inspection area and perform unilateral buffering on the area to be inspected according to the mask boundary.

[0079] Before performing multi-level screening, the prior data is first preprocessed. This preprocessing includes masking and buffering and layering. Before layering, the acquired prior data is first masked to avoid pixel mixing issues caused by spatial scale differences between the prior data and the remote sensing data collected by the remote sensing quantitative equipment to be tested.

[0080] Specifically, based on the target indicators to be tested by the remote sensing quantitative equipment, the areas to be tested and non-tested areas of the prior data are determined. The non-tested areas are masked, and the areas to be tested are buffered on one side based on the mask boundaries. In this embodiment, the width of the buffer is twice the spatial resolution of the remote sensing quantitative equipment to be tested. For example, if the spatial resolution of the remote sensing quantitative equipment to be tested is 10m, then the width of the buffer is two pixels, i.e., 20m.

[0081] Taking the above-mentioned layered processing of NDVI data as an example, a masking process is performed on the region of NDVI < 0.3 in the prior data, i.e., Class 0. The mask boundary is used as the basis for single-sided buffering. On this basis, when performing layered processing on the prior data after masking and buffering, the NDVI in the range of [0.3, 0.4) is divided into levels according to the gradient value of 0.1, and the NDVI is divided into 7 levels (Class 1 to Class 7, excluding Class 0). According to the calculated NDVI value of the prior data, the prior data is divided into multiple layers. Based on the result of the layered processing, the prior data is subjected to multi-level filtering processing based on grid, which will not be elaborated here.

[0082] In this embodiment, prior data is processed in layers, and based on the results of these layers, a multi-level grid-based screening is performed to determine pre-selected sampling units in each layer. Based on these pre-selected sampling units, the spatial layout of the sampling units allocated in each layer is optimized to determine the final target sampling units used for sampling. Sample data obtained by sampling based on the spatial location of these target sampling units can comprehensively consider the heterogeneity of the sample space and the spatial scale of the remote sensing quantitative equipment under test, ensuring the spatial homogeneity of the sampling units and thus improving the reliability of the authenticity verification results based on these sampling units.

[0083] Furthermore, by using a mask buffering process, the problem of mixed pixels caused by spatial scale differences between prior data and the remote sensing quantitative equipment to be tested is avoided, ensuring the accuracy of the sampling location for obtaining remote sensing data based on prior data.

[0084] In one embodiment, step 300 specifically includes:

[0085] Step 301: Calculate the first screening parameter in each grid of the screening grid based on the multi-layered data;

[0086] Step 302: Obtain the first parameter range, and based on the first filtering parameter and the first parameter range, filter out the first grid set from each grid of the filtering grid;

[0087] Step 303: Based on the multi-layered data, calculate the second filtering parameters within a preset grid range, centered on each grid in the first grid set;

[0088] Step 304: Obtain the second parameter range, and filter out the second grid set from the first grid set according to the second filtering parameter and the second parameter range;

[0089] Step 305: Determine the pre-selected sampling units in each layer of the multi-layered data according to each grid in the second grid set, wherein the first screening parameter and the second screening parameter include the coefficient of variation and the cell coverage ratio, and each grid of the screening grid is a basic sampling unit.

[0090] When performing multi-level filtering on multi-layered data based on the constructed filtering grid, the first filtering parameter of each grid in the filtering grid is calculated according to each layer of the multi-layered data. This first filtering parameter includes the coefficient of variation and cell coverage ratio of each grid, where the cell coverage ratio can be represented by the coverage ratio of a certain cell in each layer. The range of the first parameter is obtained. Based on the calculated filtering parameters and the obtained parameter range, the grids of the filtering grid are filtered to obtain the first grid set. Based on the multi-level filtering rules of the established grid, all grids in the established filtering grid are first used as candidate sampling units. The coefficient of variation and the coverage ratio of a certain type of cell for each grid are calculated. Then, grids that meet the conditions shown in Formula 2 are recorded and marked as Grid2:

[0091] coef1<=Quantile(coef1,20%)&Perc1>Quantile(Perc1,95%) (2)

[0092] Wherein, Quantile represents the quantile function, i.e. the parameter range, coef1 is the coefficient of variation of the filter grid, and Perc1 is the proportion of a certain type of cell covered in the filter grid. In Formula 2, the parameter range of the coefficient of variation is 20%, and the parameter range of the cell coverage proportion is 95%. The filter condition shown in Formula 2 is specifically expressed as selecting each grid with the first 20% of the coefficient of variation and the last 95% of the cell coverage proportion according to the percentile.

[0093] The selection of grids is based on the coefficient of variation and pixel coverage ratio of each grid, ensuring that the selected grids have good spatial homogeneity. Based on the first set of selected grids, and according to multi-layer data, a second selection parameter is calculated within a preset grid range, centered on each grid in the first set. This ensures that the neighboring pixels of each grid in the first set also have good spatial homogeneity, reducing verification bias caused by geometric errors of remote sensing quantitative equipment. Specifically, with Grid2 as the center, the coefficient of variation (coef2) and the coverage ratio (Perc2) of a certain pixel within the preset grid range are calculated. The preset grid range is, for example, 3Res_grid*3Res_grid, where Res_grid is the spatial resolution of the selected grid. Then, the grids that meet the selection conditions shown in Formula 3 are recorded and marked as Grid3:

[0094] coef2<=Quantile(coef2,30%)&Perc12>Quantile(Perc2,60%) (3)

[0095] Formula 3 specifically represents selecting the top 30% of the coefficients of variation within the 3Res_grid*3Res_grid range from the first grid set, and the grids whose coverage of a certain cell within this spatial range is the bottom 60%.

[0096] Based on the selected grid Grid3, the pre-selected sampling units in each layer of the multi-layered data are determined. It is understood that the above parameter ranges, filtering conditions, etc. are only used for illustrative purposes. In actual applications, the parameter values ​​can be set as needed. Furthermore, the number of filtering layers and the filtering process of multi-level filtering are not limited to those shown in this embodiment. The specific number of filtering layers, the filtering conditions of each layer, and the parameter range are all adjustable parameters and are not specifically limited here.

[0097] Further, in step 400, the target sampling unit is determined based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling units, specifically including:

[0098] Step 401: Spatially balance the number of samples allocated to each layer of the multi-layered data to obtain the set of sampling units in each layer of the multi-layered data.

[0099] Step 402: Calculate the Euclidean distance between each pair of sampling units in the sampling unit set, and filter out sampling unit pairs whose Euclidean distance is less than a preset distance threshold;

[0100] Step 403: Based on the coefficient of variation in the first screening parameters, determine the first sampling unit with a smaller coefficient of variation and the second sampling unit with a larger coefficient of variation in the sampling unit pair;

[0101] Step 404: Using the first sampling unit as the center, search for alternative sampling units for the second sampling unit based on the pre-selected sampling units;

[0102] Step 405: Replace the second sampling unit with the alternative sampling unit to obtain the target sampling unit.

[0103] After selecting pre-selected sampling units for each stratum and allocating the total number of samples to each stratum, the spatial layout of the sampling units allocated in each stratum is optimized based on the number of samples allocated to each stratum and the pre-selected sampling units. Specifically, the number of samples in each stratum is first spatially balanced to determine the spatial layout of the sampling units after balanced allocation, resulting in a set of sampling units for each stratum. The Euclidean distance between any two sampling units in the same stratum is calculated, and the sampling units in the set of sampling units are optimized based on this Euclidean distance, thereby optimizing the spatial layout of the sampling units. A spatially balanced sampling model is used to determine the initial spatial layout of the basic sampling units in each stratum. Based on this initial spatial layout, sampling units where the Euclidean distance between any two sample points is lower than a preset distance threshold are optimized to avoid sampling redundancy, resulting in the final spatial layout of the basic sampling units.

[0104] When optimizing sampling units whose Euclidean distance is lower than a preset distance threshold, the Euclidean distance between each pair of sampling units in the sampling unit set is first calculated. Pairs of sampling units whose Euclidean distance is less than the preset distance threshold are selected; this preset distance threshold is, for example, twice the Res_grid. By comparing their coefficients of variation, the first and second sampling units with smaller coefficients of variation in each pair are identified. Then, using the first sampling unit with the smaller coefficient of variation as the center, a replacement sampling unit for the second sampling unit is searched on Grid3. The second sampling unit with the larger coefficient of variation in the sampling unit pair is replaced with the found replacement sampling unit to obtain the target sampling unit.

[0105] Furthermore, step 404 may also include:

[0106] Step 4041: Obtain search parameters, including search distance and search step size;

[0107] Step 4042: Using the first sampling unit as the center, search the pre-selected sampling units outward from the search distance according to the search step size;

[0108] Step 4043: When the pre-selected sampling unit is found, the pre-selected sampling unit with the smallest coefficient of variation is selected from the pre-selected sampling units found according to the coefficient of variation in the first screening parameter as the alternative sampling unit for the second sampling unit.

[0109] When searching for alternative sampling units for the second sampling unit, search parameters are obtained, including search step size and search distance. Centered on the first sampling unit with the smallest coefficient of variation in the sampling unit alignment, the search proceeds outwards at the search distance according to the search step size, progressively searching for pre-selected sampling units (Grid3) as alternative sampling units for the second sampling unit with the largest coefficient of variation in the sampling unit alignment. Further, when multiple Grid3s are found in a certain step, the Grid3 with the smallest coefficient of variation is selected as the alternative sampling unit for the second sampling unit based on the calculated coefficients of variation of each grid. For example, taking a search distance of 3 times the Res_grid and a search step size of 1 Res_grid as an example, centered on the first sampling unit, the search proceeds outwards within a range of 3 times the Res_grid, progressively searching for Grid3s with a search step size of 1 Res_grid. When multiple Grid3 grids satisfying the conditions exist within a certain search step size range, the grid with the smallest coefficient of variation is selected as the alternative sampling unit. After optimizing each sampling unit pair, the final target sampling unit and its spatial layout are obtained. The target sampling units selected in each layer are merged using a spatial joint method to finally obtain the spatial location of the sampling units in the experimental area corresponding to the entire remote sensing data.

[0110] Furthermore, referring to Figure 2 , Figure 2 This is one of the flowcharts illustrating the sampling location determination method provided in this embodiment. Figure 2 The required data includes a pre-set number of sampling points (total number of samples), high spatial resolution prior data relative to the remote sensing quantitative device under test, and remote sensing data acquired by the device. After data acquisition, preprocessing is required, including masking and buffering of the prior data and layering. Based on the layering results, the prior data undergoes grid-based multi-level sampling unit selection. Figure 2The paper describes a two-level screening process for the selection grid. First, all grids in the established selection grid are used as the first-level selection grid (Grid1). The heterogeneity of Grid1 is determined based on the selection parameters of each grid, and screening is performed based on this heterogeneity to ensure that the selected grid (Grid2) has good spatial homogeneity. Then, Grid2 is used as the second-level selection grid. The selection parameters of neighboring pixels of Grid2 are calculated, and the heterogeneity of the neighboring region of Grid2 is determined based on the calculated selection parameters for a second-level screening, ensuring that the neighboring pixels of the selected grid (Grid3) also have good spatial homogeneity, reducing verification bias caused by geometric errors of the remote sensing quantitative equipment. The heterogeneity of the grid is determined based on the grid's selection parameters, namely the coefficient of variation and pixel coverage ratio. Finally, the total number of samples is allocated to each layer, and spatially balanced sampling is performed within each layer to achieve spatial distribution of sampling units across the layers. Calculate the Euclidean distance between each pair of sampling units in the same layer. For sampling unit pairs whose Euclidean distance is less than a preset distance threshold, remove the sampling unit with a larger coefficient of variation. Then, take the sampling unit with a smaller coefficient of variation as the center, use Grid3 as the third layer of filtering grid, and search for alternative sampling units on Grid3 to realize the spatial layout of sampling points allocated to each layer and optimize it based on Euclidean distance.

[0111] In this embodiment, by performing multi-level screening of sampling units, the selected sampling units and their neighboring pixels are ensured to have good spatial homogeneity, reducing verification bias. Simultaneously, spatial layout optimization based on Euclidean distance is performed on the sampling units allocated to each layer using spatially balanced sampling, avoiding sampling redundancy and further improving the reliability of the authenticity verification results based on the selected sampling units.

[0112] The sampling location determination device provided by the present invention is described below. The sampling location determination device described below and the sampling location determination method described above can be referred to in correspondence.

[0113] Reference Figure 3 The sampling location determination device provided in this embodiment of the invention includes:

[0114] The data acquisition module 10 is used to acquire remote sensing data collected by the remote sensing quantitative device to be verified, as well as prior data corresponding to the remote sensing data, wherein the spatial resolution of the prior data is higher than that of the remote sensing data.

[0115] The hierarchical processing module 20 is used to perform hierarchical processing on the prior data to obtain multi-layered data corresponding to the prior data.

[0116] The grid filtering module 30 is used to construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data.

[0117] The filtering and optimization module 40 is used to obtain the total number of samples and allocate the total number of samples to the multi-layered data, and determine the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling unit;

[0118] The location determination module 50 is used to determine the sampling location of the remote sensing data based on the spatial location of the target sampling unit.

[0119] In one embodiment, the screening optimization module 30 is further configured to:

[0120] Based on the multi-layered data, calculate the first screening parameter in each grid of the screening grid;

[0121] Obtain a first parameter range, and based on the first filtering parameter and the first parameter range, filter out a first set of grids from each grid of the filtering grid;

[0122] Based on the multi-layered data, calculate the second filtering parameters within a preset grid range, centered on each grid in the first grid set;

[0123] Obtain the second parameter range, and based on the second filtering parameter and the second parameter range, filter out the second grid set from the first grid set;

[0124] The pre-selected sampling units in each layer of the multi-layered data are determined based on each grid in the second grid set, wherein the first filtering parameter and the second filtering parameter include the coefficient of variation and the cell coverage ratio, and each grid of the filtering grid is a basic sampling unit.

[0125] In one embodiment, the screening optimization module 40 is further configured to:

[0126] The number of samples allocated to each layer of the multi-layered data is spatially balanced to obtain the set of sampling units in each layer of the multi-layered data;

[0127] Calculate the Euclidean distance between each pair of sampling units in the sampling unit set, and filter out sampling unit pairs whose Euclidean distance is less than a preset distance threshold;

[0128] Based on the coefficient of variation in the first screening parameters, the sampling unit is determined to be a first sampling unit with a smaller coefficient of variation and a second sampling unit with a larger coefficient of variation.

[0129] Centered on the first sampling unit, search for alternative sampling units for the second sampling unit based on the pre-selected sampling units;

[0130] The second sampling unit is replaced by the alternative sampling unit to obtain the target sampling unit.

[0131] In one embodiment, the screening optimization module 40 is further configured to:

[0132] Obtain search parameters, including search distance and search step size;

[0133] Centered on the first sampling unit, the pre-selected sampling units are searched outwards from the search distance according to the search step size;

[0134] When the pre-selected sampling unit is found, the pre-selected sampling unit with the smallest coefficient of variation is selected from the pre-selected sampling units found, based on the coefficient of variation in the first screening parameter, as the replacement sampling unit for the second sampling unit.

[0135] In one embodiment, the position determination module 50 is further configured to:

[0136] Spatial joint processing is performed on the target sampling unit to determine the spatial position of the target sampling unit in the prior data;

[0137] The spatial location is mapped onto the remote sensing data to determine the sampling location of the remote sensing data.

[0138] In one embodiment, the hierarchical processing module 20 is further configured to:

[0139] The gradient values ​​of the target indicators to be tested by the remote sensing quantitative device are obtained, and the target indicators are divided into multiple levels based on the gradient values;

[0140] Calculate the target index of the prior data, and perform stratification processing on the prior data according to the level corresponding to the target index of the prior data to obtain multi-layered data corresponding to the prior data.

[0141] In one embodiment, the sampling location determination device further includes a preprocessing module, used for:

[0142] Based on the target indicators to be tested by the remote sensing quantitative equipment, the areas to be tested and the areas not to be tested in the prior data are determined.

[0143] The non-inspection area is masked, and the area to be inspected is buffered on one side according to the mask boundary.

[0144] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a sampling position determination method, which includes:

[0145] The remote sensing data collected by the remote sensing quantitative device to be verified is obtained, as well as the prior data corresponding to the remote sensing data, wherein the spatial resolution of the prior data is higher than that of the remote sensing data.

[0146] The prior data is processed into layers to obtain multi-layered data corresponding to the prior data;

[0147] Construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data.

[0148] Obtain the total number of samples and allocate the total number of samples to the multi-layered data. Determine the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling unit.

[0149] The sampling location of the remote sensing data is determined based on the spatial location of the target sampling unit.

[0150] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0151] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute the sampling position determination method provided by the above methods, the method comprising:

[0152] The remote sensing data collected by the remote sensing quantitative device to be verified is obtained, as well as the prior data corresponding to the remote sensing data, wherein the spatial resolution of the prior data is higher than that of the remote sensing data.

[0153] The prior data is processed into layers to obtain multi-layered data corresponding to the prior data;

[0154] Construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data.

[0155] Obtain the total number of samples and allocate the total number of samples to the multi-layered data. Determine the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling unit.

[0156] The sampling location of the remote sensing data is determined based on the spatial location of the target sampling unit.

[0157] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the sampling position determination method provided by the methods described above, the method comprising:

[0158] The remote sensing data collected by the remote sensing quantitative device to be verified is obtained, as well as the prior data corresponding to the remote sensing data, wherein the spatial resolution of the prior data is higher than that of the remote sensing data.

[0159] The prior data is processed into layers to obtain multi-layered data corresponding to the prior data;

[0160] Construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data.

[0161] Obtain the total number of samples and allocate the total number of samples to the multi-layered data. Determine the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling unit.

[0162] The sampling location of the remote sensing data is determined based on the spatial location of the target sampling unit.

[0163] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0164] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining sampling location, applied to a remote sensing quantitative device to verify authenticity, characterized in that, The method includes the following steps: The remote sensing data collected by the remote sensing quantitative device to be verified is obtained, as well as the prior data corresponding to the remote sensing data. The prior data obtained is the same phenological period as the remote sensing data in time, can reflect the spatiotemporal characteristics of the parameter to be verified, and has a higher spatial resolution than the remote sensing data of the remote sensing quantitative device. The prior data is processed into layers to obtain multi-layered data corresponding to the prior data; Construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data. Obtain the total number of samples and allocate the total number of samples to the multi-layered data. Determine the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling unit. The sampling location of the remote sensing data is determined based on the spatial location of the target sampling unit.

2. The sampling location determination method according to claim 1, characterized in that, The step of performing multi-level filtering on the multi-layered data based on the filtering grid to determine the pre-selected sampling units in each layer of the multi-layered data includes: Based on the multi-layered data, calculate the first screening parameter in each grid of the screening grid; Obtain a first parameter range, and based on the first filtering parameter and the first parameter range, filter out a first set of grids from each grid of the filtering grid; Based on the multi-layered data, calculate the second filtering parameters within a preset grid range, centered on each grid in the first grid set; Obtain the second parameter range, and based on the second filtering parameter and the second parameter range, filter out the second grid set from the first grid set; The pre-selected sampling units in each layer of the multi-layered data are determined based on each grid in the second grid set, wherein the first filtering parameter and the second filtering parameter include the coefficient of variation and the cell coverage ratio, and each grid of the filtering grid is a basic sampling unit.

3. The sampling location determination method according to claim 2, characterized in that, The step of determining the target sampling unit based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling units includes: The number of samples allocated to each layer of the multi-layered data is spatially balanced to obtain the set of sampling units in each layer of the multi-layered data; Calculate the Euclidean distance between each pair of sampling units in the sampling unit set, and filter out sampling unit pairs whose Euclidean distance is less than a preset distance threshold; Based on the coefficient of variation in the first screening parameters, the sampling unit is determined to be a first sampling unit with a smaller coefficient of variation and a second sampling unit with a larger coefficient of variation. Centered on the first sampling unit, search for alternative sampling units for the second sampling unit based on the pre-selected sampling units; The second sampling unit is replaced by the alternative sampling unit to obtain the target sampling unit.

4. The sampling location determination method according to claim 3, characterized in that, The step of searching for alternative sampling units for the second sampling unit based on the pre-selected sampling units, with the first sampling unit as the center, includes: Obtain search parameters, including search distance and search step size; Centered on the first sampling unit, the pre-selected sampling units are searched outwards from the search distance according to the search step size; When the pre-selected sampling unit is found, the pre-selected sampling unit with the smallest coefficient of variation is selected from the pre-selected sampling units found, based on the coefficient of variation in the first screening parameter, as the replacement sampling unit for the second sampling unit.

5. The sampling location determination method according to claim 1, characterized in that, The step of determining the sampling location of the remote sensing data based on the spatial location of the target sampling unit includes: Spatial joint processing is performed on the target sampling unit to determine the spatial position of the target sampling unit in the prior data; The spatial location is mapped onto the remote sensing data to determine the sampling location of the remote sensing data.

6. The sampling location determination method according to claim 1, characterized in that, The step of performing stratified processing on the prior data to obtain multi-stratified data corresponding to the prior data includes: The gradient values ​​of the target indicators to be tested by the remote sensing quantitative equipment are obtained, and the target indicators are divided into multiple levels based on the gradient values; Calculate the target index of the prior data, and perform stratification processing on the prior data according to the level corresponding to the target index of the prior data to obtain multi-layered data corresponding to the prior data.

7. The sampling location determination method according to claim 6, characterized in that, Before the step of performing hierarchical processing on the prior numbers, the method further includes: Based on the target indicators to be tested by the remote sensing quantitative equipment, the areas to be tested and the areas not to be tested in the prior data are determined. The non-inspection area is masked, and the area to be inspected is buffered on one side according to the mask boundary.

8. A sampling location determination device, characterized in that, include: The data acquisition module is used to acquire remote sensing data collected by the remote sensing quantitative device to be verified, as well as the prior data corresponding to the remote sensing data. The acquired prior data is the same phenological period as the remote sensing data, can reflect the spatiotemporal characteristics of the parameter to be verified, and has a higher spatial resolution than the remote sensing data of the remote sensing quantitative device. A hierarchical processing module is used to perform hierarchical processing on the prior data to obtain multi-layered data corresponding to the prior data. The grid filtering module is used to construct a filtering grid for the multi-layered data, perform multi-level filtering on the multi-layered data based on the filtering grid, and determine the pre-selected sampling units in each layer of the multi-layered data. The filtering and optimization module is used to obtain the total number of samples and allocate the total number of samples to the multi-layered data. Based on the number of samples allocated to each layer of the multi-layered data and the pre-selected sampling units, the target sampling unit is determined. The location determination module is used to determine the sampling location of the remote sensing data based on the spatial location of the target sampling unit.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the sampling position determination method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the sampling position determination method as described in any one of claims 1 to 7.

Images