A method and device for automatically refining the fuzzy positions of historical soil survey sampling points

Abstract

The present invention discloses a method and device for automatically refining the fuzzy positions of historical soil survey sampling points. The method mainly includes: performing structured processing on the scene description text of the target sampling point and the reference sampling point to obtain location information and geographic scene information; loading the geographic element vector layer to generate a geographic scene unit layer; eliminating scene units that are not adjacent to the sampling point through preliminary data screening; using the reference sampling point positioning data as an important reference standard, using the entropy method to obtain the geographic scene feature importance weight; matching the target sampling point with the scene unit one by one, and calculating the similarity between the target sampling point and each scene unit in the geographic scene unit layer based on the membership function and the feature importance weight, and then finding the scene unit with the highest similarity, and assigning the precise position of the target sampling point to the center point position of the scene unit. The present invention is time-saving and highly accurate.

Description

Technical Field

[0001] The present invention relates to the field of geographic information technology, and in particular to a method and device for automatically refining the fuzzy positions of historical soil survey sample points. Background Art

[0002] While current field sampling data often includes precise spatial location information, much historical sample survey data often only provides simple spatial location descriptions without precise spatial location information, severely limiting the usefulness of these survey data. Given this crude sample location description, how to spatially constrain these fuzzy locations using geographic information such as the sample's parent material and elevation, and refine these fuzzy locations to a relatively accurate position, is of great research significance and application value.

[0003] Currently, the main approach to addressing this issue is to first standardize place names, rewriting them into the format of "xx Province, xx City, xx County (City / District), xx Town, xx Village, xx"; then, utilize network coding positioning to obtain longitude and latitude coordinates; and finally, optimize the location by combining topological and azimuthal relationships, followed by manual identification and adjustment. However, both the standardization of place names and the subsequent manual identification of the locations are time-consuming and labor-intensive. Furthermore, due to the manual identification process, the objectivity and reliability of the manually adjusted data are difficult to guarantee. Summary of the Invention

[0004] In view of the problems existing in the prior art, the purpose of the present invention is to provide a method and equipment for automatically refining the fuzzy positions of historical soil survey sample points with short time consumption and high accuracy.

[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0006] A method for automatically refining the fuzzy positions of historical soil survey sample points comprises the following steps:

[0007] (1) All soil survey sampling points to be precisely located in the target area are taken as target sampling points, and all soil survey sampling points that have been precisely located in the target area are taken as reference sampling points. The scene description texts of the target sampling points and the reference sampling points are structured to obtain the rough positioning position and geographic scene feature value of the target sampling points, as well as the precise position and geographic scene feature value of the reference sampling points;

[0008] (2) Generate the target sample point distribution layer P1 based on the coarse positioning position, and generate the reference sample point distribution layer P2 based on the precise position of the reference sample point;

[0009] (3) Obtaining a scene unit distribution map of all geographic scene features of the target area, and overlaying the scene unit distribution maps of all geographic scene features of the target area to generate a geographic scene unit layer S1;

[0010] (4) Combined with the reference sample point distribution layer P2, calculate the similarity between each reference sample point and the geographic scene feature value of its corresponding scene unit on the geographic scene unit layer S1, and calculate the feature importance weight of each geographic scene feature based on the similarity, where the lower the similarity corresponding to the geographic scene feature, the greater the feature importance weight;

[0011] (5) Based on the target sample point distribution layer P1, the non-adjacent units in the geographic scene unit layer S1 are eliminated to generate the geographic scene unit layer S2;

[0012] (6) Combine the target sample point distribution layer P1 and the geographic scene unit layer S2 for analysis to obtain the geographic scene feature value of the scene unit corresponding to each target sample point on the geographic scene unit layer S2, as well as the geographic scene feature value of the adjacent scene unit;

[0013] (7) Calculate the similarity between each geographic scene feature value of each target sample point and the corresponding geographic scene feature value of the corresponding scene unit, and use the feature importance weight as the weight to calculate the weighted sum of the similarities between all geographic scene feature values ​​as the similarity sim0 between each target sample point and the corresponding scene unit;

[0014] (8) If the similarity sim0 is greater than the threshold δ, the precise position of the target sample point is assigned to the center point position of the corresponding scene unit. Otherwise, the similarity between the target sample point and all adjacent scene units is calculated according to step (7), and the maximum value is obtained. If the maximum value is greater than the threshold δ, the precise position of the target sample point is assigned to the center point position of the adjacent scene unit corresponding to the maximum value.

[0015] (9) If the similarity between the target sample point and the corresponding scene unit and the adjacent scene unit is less than the threshold δ, the similarity between the target sample point and all the scene units in the administrative area where the target sample point is located on the geographic scene unit layer S2 is calculated, and the precise position of the target sample point is assigned to the center point position of the scene unit with the greatest similarity.

[0016] Furthermore, step (1) specifically includes:

[0017] (1-1) Obtain scene description texts for all target samples and reference samples;

[0018] (1-2) Recognizing the scene description text by using a preset regular expression to obtain the location description text and geographic scene feature values ​​of the target sample point, as well as the precise location and geographic scene feature values ​​of the reference sample point;

[0019] (1-3) The location description text of the target sample point is processed according to the administrative address format to obtain the administrative location of the target sample point, and rough positioning is performed based on the administrative location to obtain the rough positioning location.

[0020] Furthermore, step (2) specifically includes:

[0021] (2-1) Generate target sample point distribution layer P1 based on the coarse positioning positions of all target sample points;

[0022] (2-2) Generate a reference sample point distribution layer P2 based on the precise locations of all reference sample points.

[0023] Furthermore, step (4) specifically includes:

[0024] (4-1) According to the reference sample point distribution layer P2, obtain the scene unit corresponding to the reference sample point on the geographic scene unit layer S1;

[0025] (4-2) For the numerical type geographic scene feature value of the reference sample point, using a membership function that is suitable for the change characteristics of the corresponding geographic scene feature, calculate the similarity between the geographic scene feature value and the corresponding geographic scene feature value of the scene unit corresponding to the reference sample point as the membership value of the geographic scene feature value;

[0026] (4-3) for the enumerated type geographic scene feature value of the reference sample point, obtaining, according to a preset mapping rule, a similarity between the geographic scene feature value and the corresponding geographic scene feature value of the scene unit corresponding to the reference sample point, as the membership value of the geographic scene feature value, wherein the preset mapping rule is used to indicate the mapping relationship between the geographic scene feature value and the similarity;

[0027] (4-4) Calculate the feature importance weight of each geographic scene feature based on the membership values ​​of all geographic scene feature values ​​of the reference sample points, wherein the lower the membership value of the geographic scene feature, the greater the feature importance weight.

[0028] Furthermore, step (4-2) specifically includes:

[0029] (4-2-1) For geographic scene features of numerical type whose data variation does not exceed the preset range, a triangular membership function or a trapezoidal membership function is used to calculate the similarity between the corresponding geographic scene feature values ​​as the membership value of the corresponding geographic scene feature value;

[0030] (4-2-2) For geographic scene features of numerical type whose data variation exceeds the preset range, any function among the polynomial membership function, Gaussian membership function, and root membership function is used to calculate the similarity between the corresponding geographic scene feature values ​​as the membership value of the corresponding geographic scene feature value.

[0031] Furthermore, the specific formula for calculating similarity using the triangular membership function is:

[0032] μ=1-|xp-xs| / (xs-a)

[0033] Where μ is the similarity, xp is the geographic scene feature value of the reference sample point, xs is the geographic scene feature value of the scene unit corresponding to the reference sample point on the geographic scene unit layer S1, and a is a variable parameter;

[0034] The specific formula for calculating similarity using the trapezoidal membership function is:

[0035]

[0036] Where x′=xs±b, b is the neighborhood radius of the trapezoidal maximum value, and c is a variable parameter;

[0037] The specific formula for calculating similarity using the Gaussian membership function is:

[0038]

[0039] Where σ is the standard deviation of the membership function;

[0040] The specific formula for calculating similarity using the polynomial membership function is:

[0041] μ=a0+a1|xp-xs|+a2x 2 |xp-xs| 2 +...+at|xp-xs| t

[0042] Where a0, a1, ..., at are the coefficients of the polynomial;

[0043] The specific formula for calculating similarity using the root membership function is:

[0044]

[0045] Where d is a variable parameter.

[0046] Furthermore, step (5) specifically includes:

[0047] (5-1) Filter out the township-level administrative units where each soil survey sampling point is located to generate the township layer C;

[0048] (5-2) On the target sampling point distribution layer P1, a number of circular adjacent areas are divided with each soil survey sampling point as the center and a preset distance as the radius. The areas outside the circular adjacent areas on the target sampling point distribution layer P1 are deleted to generate a buffer layer B;

[0049] (5-3) Perform an intersection operation on the geographic scene unit layer S1, the township layer C, and the buffer layer B to generate the geographic scene unit layer S2.

[0050] Furthermore, step (6) specifically includes:

[0051] (6-1) Solve the topological relationship between the target sample point distribution layer P1 and the geographic scene unit layer S2 to obtain the corresponding scene unit of each target sample point on the geographic scene unit layer S2;

[0052] (6-2) extracting all geographic scene feature values ​​of the corresponding scene unit of each target sample point on the geographic scene unit layer S2;

[0053] (6-3) By solving the topological relationship, all scene units adjacent to the scene unit corresponding to each target sample point are obtained, and all geographic scene feature values ​​of these adjacent scene units are extracted.

[0054] Furthermore, step (7) specifically includes:

[0055] (7-1) For different geographic scene feature values, select the membership function consistent with step (4), calculate the similarity between each geographic scene feature value of each target sample point and the corresponding geographic scene feature value of the corresponding scene unit on the geographic scene unit layer S2, and use it as the membership value of the geographic scene feature corresponding to the target sample point;

[0056] (7-2) Taking the feature importance weight of the geographic scene feature as the corresponding weight, calculate the weighted sum of the membership value of each geographic scene feature of the target sample point as the similarity sim0 between each target sample point and the corresponding scene unit.

[0057] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above method.

[0058] Compared with the prior art, the present invention has the following beneficial effects: the present invention realizes a method for automatically refining the fuzzy position of historical soil survey sample points through processing steps such as data structured processing, coarse positioning based on location description information, acquisition of geographic environment information based on geographic scene units, geographic scene similarity calculation, acquisition of the best matching geographic scene unit, and sample point positioning. The present invention can greatly reduce the cost of manual intervention and effectively improve processing efficiency while achieving positioning results with an accuracy similar to or even better than that obtained by manual processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Flowchart of the method for automatically refining the fuzzy positions of historical soil survey sample points provided in this embodiment;

[0060] Figure 2 This is a distribution diagram of the PR samples, PA samples, and PT samples used in this embodiment;

[0061] Figure 3 This is the distribution layer of subcategories required for overlay analysis in this embodiment;

[0062] Figure 4 This is the distribution layer of the mapped parent material required for the overlay analysis in this embodiment;

[0063] Figure 5 The distribution layer of topography required for the overlay analysis in this embodiment;

[0064] Figure 6 The scene unit distribution layer S1 is formed by overlay analysis in this embodiment;

[0065] Figure 7 This is the township distribution layer C obtained based on the sample point PR in this embodiment;

[0066] Figure 8 This is a buffer distribution layer B with a radius of 5 km set based on the sample point PR in this embodiment;

[0067] Figure 9 The search range distribution layer BC is generated based on the intersection operation of layers C and B in this embodiment;

[0068] Figure 10 The experimental scene unit distribution layer S2 is selected based on the intersection operation of the layers S and BC in this embodiment;

[0069] Figure 11 : This is a distribution diagram of the coordinates PRC obtained by solving the PR sample point and the coordinates PR of the coarse positioning in this embodiment;

[0070] Figure 12This is a distribution diagram of the coordinates PTC obtained by solving the PT sample points in this embodiment, the coordinates PT of the rough positioning, and the precise position coordinates RPT. DETAILED DESCRIPTION

[0071] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0072] The experimental data for this example are derived from the Second National Soil Survey data for Jiangxi Province and the Soil System Survey data for Jiangxi Province. The geographic scene features required for this example are six attributes: parent material, soil subtype, topography, temperature, precipitation, and altitude. The parent material, soil subtype, and topography data are vector data, while the remaining three are raster data. The purpose of the experiment is to obtain the matching of scene feature information between the target sample point and the scene unit, and to comprehensively analyze and determine the coordinates of the fuzzy position to improve the spatial positioning accuracy and data usability of the historical field sample survey data, facilitating subsequent comparison of ancient and modern soil data.

[0073] The method for automatically refining the fuzzy position of historical soil survey sample points provided by the embodiment of the present invention is as follows: Figure 1 As shown, the specific steps include:

[0074] (1) All soil survey sample points to be precisely located in the target area are taken as target sample points, and all soil survey sample points that have been precisely located in the target area are taken as reference sample points. The scene description texts of the target sample points and the reference sample points are structured to obtain the coarse positioning position and geographic scene feature values ​​of the target sample points, as well as the precise position and geographic scene feature values ​​of the reference sample points.

[0075] This step specifically includes:

[0076] (1-1) Obtain scene description texts for all target samples and reference samples;

[0077] (1-2) The scene description text is identified by using a preset regular expression to obtain the location description text and geographic scene feature values ​​of the target sample point, as well as the precise location and geographic scene feature values ​​of the reference sample point; wherein the regular expression can be set according to the text format, and the location description text and each geographic scene feature value of the target sample point can be identified from the scene description text by using the set regular expression. The geographic scene features specifically include parent material, soil subtype, topography, temperature, precipitation, and altitude.

[0078] (1-3) The location description text of the target sample point is processed into an administrative address format to obtain the administrative location of the target sample point. Coarse positioning is performed based on the administrative location to obtain a coarse positioning position. The administrative location is specifically in the form of xx village xx town xx county (city / district) xx province xx city xx. Coarse positioning of the administrative location based on spatial coding can obtain a roughly estimated latitude and longitude coordinate.

[0079] In this embodiment, the target soil survey sample points to be accurately located are derived from the second national soil census data of Jiangxi Province, totaling 77, and the target soil survey sample points to be accurately located are defined as target sample points PR. The soil survey sample points that have been accurately located mainly come from the soil system survey data of Jiangxi Province, totaling 69, and the soil survey sample points that have been accurately located are defined as reference sample points PA. This type of sample point has precise geographic coordinates and is mainly used to obtain weights. The sample points PT used to verify the accuracy mainly come from the soil system survey data of Jiangxi Province, totaling 50. This type of sample point has precisely located coordinates and can be used for later inspection. During the verification process, the sample point is first roughly located and then brought into the present method for inspection. The rough positioning position and geographic scene feature values ​​of the target sample point PR are shown in Table 1, the precise position and geographic scene feature values ​​of the reference sample point PA are shown in Table 2, and the precise position and geographic scene feature values ​​of the verification accuracy sample point PT are shown in Table 3.

[0080] Table 1 Target sample point information

[0081]

[0082] Table 2 Reference sample information

[0083]

[0084]

[0085] Table 3 Verification accuracy sample information

[0086]

[0087] (2) Generate the target sample point distribution layer P1 according to the coarse positioning position, and generate the reference sample point distribution layer P2 according to the precise position of the reference sample point.

[0088] This step specifically includes:

[0089] (2-1) Generate target sample point distribution layer P1 based on the coarse positioning positions of all target sample points;

[0090] (2-2) Generate a reference sample point distribution layer P2 based on the precise locations of all reference sample points.

[0091] In addition, for verification, the coarse positioning latitude and longitude coordinates of the verification sample point PT are obtained based on the network space coding service to generate the verification sample point distribution layer P3. In this embodiment, for simplicity, the target sample point distribution layer P1, the reference sample point distribution layer P2 and the verification sample point distribution layer P3 are displayed in one figure, as shown in the following figure. Figure 2 shown.

[0092] (3) Obtain a scene unit distribution map of all geographic scene features of the target area, and overlay the scene unit distribution maps of all geographic scene features of the target area to generate a geographic scene unit layer S1.

[0093] In this embodiment, the vector scene unit distribution maps of geographical scene features such as parent material, soil subtype, topography, etc. are superimposed based on GIS software, and then raster data such as precipitation, temperature, and altitude are calculated and assigned to the corresponding scene units to generate a geographical scene unit layer S1. Among them, the scene unit distribution maps of parent material, soil subtype, and topography in this embodiment are respectively as follows: Figure 3 、 Figure 4 、 Figure 5 As shown, the generated geographic scene unit layer S1 is as follows Figure 6 shown.

[0094] (4) Combined with the reference sample point distribution layer P2, calculate the similarity between each reference sample point and the geographic scene feature value of its corresponding scene unit on the geographic scene unit layer S1, and calculate the feature importance weight of each geographic scene feature based on the similarity.

[0095] This step specifically includes:

[0096] (4-1) According to the reference sample point distribution layer P2, obtain the scene unit corresponding to the reference sample point on the geographic scene unit layer S1;

[0097] (4-2) For the numerical type of geographic scene feature value of the reference sample point, such as temperature, precipitation, altitude, etc., a membership function suitable for the change characteristics of the corresponding geographic scene feature is used to calculate the similarity between the geographic scene feature value and the corresponding geographic scene feature value of the scene unit corresponding to the reference sample point, as the membership value of the geographic scene feature value;

[0098] (4-3) for a geographic scene feature value of an enumerated type of a reference sample point, such as parent material, subclass, land use type, etc., obtaining, according to a preset mapping rule, a similarity between the geographic scene feature value and a corresponding geographic scene feature value of a scene unit corresponding to the reference sample point, as a membership value of the geographic scene feature value, wherein the preset mapping rule is used to indicate a mapping relationship between the geographic scene feature value and the similarity;

[0099] (4-4) Calculate the feature importance weight of each geographic scene feature based on the membership value of all geographic scene feature values ​​of the reference sample point, wherein the lower the membership value of the geographic scene feature, the greater the feature importance weight. For example, the feature importance weight λ of each geographic scene feature can be calculated by the entropy method. i In this example, the feature importance weights of subtype, topography, parent material, temperature, precipitation, and altitude are 0.566607, 0.058110, 0.297511, 0.031473, 0, 0.013905, and 0.032395, respectively.

[0100] The geographic scene feature information of the scene units corresponding to all reference sample points on the geographic scene unit layer S1 is shown in Table 4 below.

[0101] Table 4 Geographic scene feature information of scene units corresponding to reference sample points

[0102] ID Earth system name Subclass Topography soil parent material Temperature precipitation altitude 1 Nanshengli faction Paddy soil 101 River-lake erosion / sediment 17.925 1546.921 17.27175 2 Later Zhou Dynasty Paddy soil 101 River-lake erosion / sediment 18.20219 1644.6 18.49621 3 Street Paddy soil 101 River-lake erosion / sediment 17.78078 1530.416 16.103 4 Moshan System Paddy soil 101 Quaternary sub-red sandy soil 17.94481 1571.232 24.66764 5 Under the cedar tree red soil 211 Quartzite 18.6883 1384.282 266.7122 6 Xiejiatan system Paddy soil 101 River-lake erosion / sediment 17.675 1530.262 21.87786 7 Yunqian Department paddy soil 101 Mudstone 17.91416 1567.116 26.81573 8 Yang Jian family red soil 211 other 17.42955 1432 214.9281 9 Rentian system Calcareous purple soil 101 Purple rocks 19.33748 1465.311 224.6062 10 Shaocun system red soil 101 Mudstone 17.675 1586 37.91094 … …… …… …… …… …… …… …… 69 Zuoxi System red soil 211 Quartzite 17.61543 1358.428 381.3419

[0103] Wherein, step (4-2) specifically includes:

[0104] (4-2-1) For numerical geographic scene features whose data variation does not exceed a preset range, such as temperature and precipitation, where the upper and lower limits of variation within a certain area are not large, a triangular membership function or a trapezoidal membership function is used to calculate the similarity between the corresponding geographic scene feature values, which is used as the membership value of the corresponding geographic scene feature value. In this embodiment, the triangular membership function is used for processing;

[0105] (4-2-2) For geographic scene features of numerical type whose data variation exceeds the preset range, such as altitude, mutations may occur within a certain area. However, for the case where the change in the properties of soil samples does not change linearly with increasing altitude, any function among the polynomial membership function, Gaussian membership function, and root membership function is used to calculate the similarity between the corresponding geographic scene feature values ​​as the membership value of the corresponding geographic scene feature value. In this embodiment, the root membership function is used for processing.

[0106] Among them, the specific formula for calculating similarity using the triangular membership function is:

[0107] μ=1-|xp-xs| / (xs-a)

[0108] Where μ is the similarity, xp is the geographic scene feature value of the reference sample point, xs is the geographic scene feature value of the scene unit corresponding to the reference sample point on the geographic scene unit layer S1, and a is a variable parameter;

[0109] The specific formula for calculating similarity using the trapezoidal membership function is:

[0110]

[0111] Where x′=xs±b, b is the neighborhood radius of the trapezoidal maximum value, and c is a variable parameter;

[0112] The specific formula for calculating similarity using the Gaussian membership function is:

[0113]

[0114] Where σ is the standard deviation of the membership function;

[0115] The specific formula for calculating similarity using the polynomial membership function is:

[0116] μ=a0+a1|xp-xs|+a2x 2 |xp-xs| 2 +...+at|xp-xs| t

[0117] Where a0, a1, ..., at are the coefficients of the polynomial;

[0118] The specific formula for calculating similarity using the root membership function is:

[0119]

[0120] Where d is a variable parameter.

[0121] In step (4-3), the preset mapping rule is the matching mapping relationship between each scene feature and the data. The relationship between different values ​​needs to be considered and the membership value is set according to the conformity of the data. For example, the preset mapping rule can be:

[0122] A: If the geographic scene feature value is completely consistent with the corresponding geographic scene feature value of the scene unit corresponding to the reference sample point, the membership value is set to 1;

[0123] B: If the relationship between the geographic scene feature value and the corresponding geographic scene feature value of the scene unit corresponding to the reference sample point is close, the membership value is set to 0.3-0.8. In this embodiment, in the soil parent material, the relationship between Quaternary red clay, Quaternary sub-red sandy soil and red sandstone is set to 0.5, and the relationship between other sediments with river and lake impact is also set to 0.5; in the subclass, the relationship between paddy soil, lagoon paddy soil, latent paddy soil and seepage paddy soil is set to 0.6, the relationship between red soil, red-yellow soil and red soil is set to 0.6, and the relationship between red soil and various paddy soils is set to 0.4; in the topography, the relationship between low-altitude plains and low-altitude hills, low-altitude hills and small undulating low mountains, small undulating low mountains and medium undulating low mountains, medium undulating low mountains and medium undulating medium mountains, and medium undulating medium mountains and large undulating medium mountains is set to 0.5.

[0124] C: If similar values ​​such as "Other" appear, consider the matching value between it and other values ​​and set the membership value to around 0.5;

[0125] D: When the other eigenvalues ​​are irrelevant, the membership value is set to 0.

[0126] (5) Based on the target sample point distribution layer P1, the non-adjacent units in the geographic scene unit layer S1 are eliminated to generate the geographic scene unit layer S2.

[0127] This step specifically includes:

[0128] (5-1) Filter out the township-level administrative units where each soil survey sample point is located and generate the township layer C, such as Figure 7 As shown;

[0129] (5-2) On the target sample point distribution layer P1, a number of circular adjacent areas are divided with each soil survey sample point as the center and the preset distance as the radius. The areas outside the circular adjacent areas on the target sample point distribution layer P1 are deleted to generate a buffer layer B, as shown in Figure 8 As shown; in this embodiment, the preset distance is 5km;

[0130] (5-3) Perform an intersection operation on the geographic scene unit layer S1, the township layer C, and the buffer layer B to generate the geographic scene unit layer S2. In this embodiment, first perform an intersection operation on layers C and B to generate the search layer BC ( Figure 9 ), then based on the intersection operation of layer S and BC, generate scene unit layer S2 ( Figure 10 ).

[0131] (6) Combine the target sample point distribution layer P1 and the geographic scene unit layer S2 for analysis to obtain the geographic scene feature value of the scene unit corresponding to each target sample point on the geographic scene unit layer S2, as well as the geographic scene feature value of the adjacent scene unit.

[0132] This step specifically includes:

[0133] (6-1) Solve the topological relationship between the target sample point distribution layer P1 and the geographic scene unit layer S2 to obtain the corresponding scene unit of each target sample point on the geographic scene unit layer S2;

[0134] (6-2) extracting all geographic scene feature values ​​of the corresponding scene unit of each target sample point on the geographic scene unit layer S2;

[0135] (6-3) By solving the topological relationship, all units adjacent to the scene unit corresponding to each target sample point are obtained, and all geographic scene feature values ​​of these adjacent scene units are extracted.

[0136] (7) Calculate the similarity between each geographic scene feature value of each target sample point and the corresponding geographic scene feature value of the corresponding scene unit, and use the feature importance weight as the weight to calculate the weighted sum of the similarities between all geographic scene feature values ​​as the similarity sim0 between each target sample point and the corresponding scene unit.

[0137] This step specifically includes:

[0138] (7-1) For different geographic scene feature values, select the membership function consistent with step (4), calculate the similarity between each geographic scene feature value of each target sample point and the similarity between the geographic scene feature value of the corresponding scene unit on the geographic scene unit layer S2, and use it as the membership value of the geographic scene feature corresponding to the target sample point; in this embodiment, a triangular membership function is used for temperature and precipitation, a radical membership function is used for altitude, and an enumeration membership function is used for the remaining three geographic scene features (parent material, soil subtype, and topography);

[0139] (7-2) Taking the feature importance weight of the geographic scene feature as the corresponding weight, the weighted sum of the membership value of each geographic scene feature of the target sample point is calculated according to the following formula as the similarity sim0 between each target sample point and the corresponding scene unit.

[0140]

[0141] Where μi is the membership value of the i-th geographic scene feature of the target sample point.

[0142] (8) If the similarity sim0 is greater than a threshold δ, the precise position of the target sample point is assigned to the center point of the corresponding scene unit. Otherwise, the similarities between the target sample point and all adjacent scene units are calculated according to step (7), and the maximum value is obtained. If the maximum value is greater than the threshold δ, the precise position of the target sample point is assigned to the center point of the adjacent scene unit corresponding to the maximum value. In this embodiment, the threshold δ is 0.7.

[0143] (9) If the similarity between the target sample point and the corresponding scene unit and the adjacent scene unit is less than the threshold δ, the similarity between the target sample point and all the scene units in the administrative area where the target sample point is located on the geographic scene unit layer S2 is calculated, and the precise position of the target sample point is assigned to the center point position of the scene unit with the greatest similarity.

[0144] In this embodiment, after the refined coordinate processing, the PRC and the PR distribution map of the target sample points originally roughly positioned are obtained ( Figure 11 ), the coordinate data are shown in Table 5.

[0145] Table 5 Position coordinates of target sample points before and after precise calibration

[0146] ID soil species Coarse positioning longitude Coarse positioning latitude Calibrated longitude Calibrated dimensions 1 Thick layer of black sandy red soil 117.05 27.44 116.9916292 27.36545012 2 Thick layer of sandy red soil 115.21 26.31 115.2104215 26.27451851 3 Thick layer of gray sandy soil 116.58 27.97 116.5374917 27.98822916 4 Thick layer of dark yellow sandy mud red soil 114.17 27.29 114.119533 27.30222102 5 Thin layer of gray-yellow sandy mud red soil 115.37 24.95 115.3456234 24.97460872 6 Thick layer of gray-yellow sandy soil 114.17 27.29 114.119533 27.30222102 7 Thick layer of black eel mud red soil 115.26 26.42 115.2145165 26.53554198 8 Thick layer of gray eel mud red soil 117.93 29.54 117.6758397 29.25254943 9 Thin layer of black eel mud red soil 116.46 28.29 116.5291958 28.3087536 10 Thick layer of gray eel soil 116.71 28.35 116.7089346 28.39133726 … … … … … … 77 Fully latent ash tide sandy mud field 116.66 28.76 116.6844226 28.71287956

[0147] After inspection and verification, the data from the second soil survey in Jiangxi Province processed by the method of the present invention has a good precision optimization effect, which is close to the accuracy of manual positioning, and even better than manual calibration. Specifically, it includes:

[0148] a) Comparative evaluation based on the environmental similarity between the target sample point PR and the scene unit in which it is located. By comparing the similarity between the calibrated unit and the initial scene unit, there is a significant improvement, and the environmental similarity between 60% of the adjusted sampling points and the unit in which they are located is greater than the set threshold. The remaining sample points are only slightly below the threshold, with only 1-2 points having relatively low values;

[0149] b) Based on the comparative evaluation of the validation sample points PT, it can be found that the similarity calculated for more than 70% of the points is greater than the threshold, and nearly 90% of the points are greater than 0.6. At the same time, by comparing with the coarse positioning coordinates and the actual positioning coordinates, 72% of the sample points meet at least one of the following conditions, that is, they meet the inspection criteria: 1) The obtained coordinates are within 3km of the actual coordinates; 2) The coordinates after correction are closer to the actual coordinates than the coarse positioning coordinates; 3) The coordinates after correction and the actual coordinates are in the same scene unit. Data such as Figure 12 As shown in Table 6. Taking into account the accuracy of the vector information used and other objective factors, it can better reflect the local geographical scene characteristics and perform relatively accurate positioning within a smaller range. The error is within the allowable range and basically meets the inspection requirements.

[0150] Table 6 Verification of inspection points

[0151]

[0152]

[0153] Different testing methods have been proven to be able to better meet people's experimental needs. They can, to a certain extent, refer to the sampling points selected for the soil survey at that time and meet the current needs of comparative analysis and research on ancient and modern soil data.

[0154] In the embodiment of the present invention, fuzzy position precision is only performed on the second soil survey data of Jiangxi Province, with the soil series survey data of Jiangxi Province as an important reference standard. The method of the present invention is also applicable to position precision methods under the constraints of geographical scene features in other fields.

[0155] An embodiment of the present invention also provides a computer device that provides services for implementing the above-mentioned method. The device may include: a memory storing a computer executable program; a processor coupled to the memory; the processor invoking the computer executable program stored in the memory 301 to execute the steps of the method described in the first embodiment. The memory may include computer system-readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the memory 301 may be used to read and write to non-removable, non-volatile magnetic media (commonly referred to as a "hard drive"). A program / utility having a set (at least one) of program modules may be stored, for example, in the memory. Such program modules include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each of these examples, or some combination thereof, may include an implementation of a network environment. The computer executable programs of the program modules generally perform the functions and / or methods described in the embodiments of the present invention. The processor executes the programs stored in the memory to execute various functional applications and data processing, such as implementing the method provided in the first embodiment of the present invention. The code of the computer executable program can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, Python, and conventional procedural programming languages ​​such as "C" language or similar programming languages.

[0156] It should be understood that the above embodiments and descriptions only describe the principles, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for automatically refining the fuzzy positions of historical soil survey sampling points, characterized in that: The steps include: (1) All soil survey sampling points to be precisely located in the target area are taken as target sampling points, and all soil survey sampling points that have been precisely located in the target area are taken as reference sampling points. The scene description texts of the target sampling points and the reference sampling points are structured to obtain the rough positioning position and geographic scene feature value of the target sampling points, as well as the precise position and geographic scene feature value of the reference sampling points; (2) Generate the target sample point distribution layer P1 based on the coarse positioning position, and generate the reference sample point distribution layer P2 based on the precise position of the reference sample point; (3) Obtaining a scene unit distribution map of all geographic scene features of the target area, and overlaying the scene unit distribution maps of all geographic scene features of the target area to generate a geographic scene unit layer S1; (4) Combined with the reference sample point distribution layer P2, calculate the similarity between each reference sample point and the geographic scene feature value of its corresponding scene unit on the geographic scene unit layer S1, and calculate the feature importance weight of each geographic scene feature based on the similarity, where the lower the similarity corresponding to the geographic scene feature, the greater the feature importance weight; (5) Based on the target sample point distribution layer P1, the non-adjacent units in the geographic scene unit layer S1 are eliminated to generate the geographic scene unit layer S2; (6) Combine the target sample point distribution layer P1 and the geographic scene unit layer S2 for analysis to obtain the geographic scene feature value of the scene unit corresponding to each target sample point on the geographic scene unit layer S2, as well as the geographic scene feature value of the adjacent scene unit; (7) Calculate the similarity between each geographic scene feature value of each target sample point and the corresponding geographic scene feature value of the corresponding scene unit, and use the feature importance weight as the weight to calculate the weighted sum of the similarities between all geographic scene feature values ​​as the similarity sim0 between each target sample point and the corresponding scene unit; (8) If the similarity sim0 is greater than the threshold δ, the precise position of the target sample point is assigned to the center point position of the corresponding scene unit. Otherwise, the similarity between the target sample point and all adjacent scene units is calculated according to step (7), and the maximum value is obtained. If the maximum value is greater than the threshold δ, the precise position of the target sample point is assigned to the center point position of the adjacent scene unit corresponding to the maximum value. (9) If the similarity between the target sample point and the corresponding scene unit and the adjacent scene unit is less than the threshold δ, the similarity between the target sample point and all the scene units in the administrative area where the target sample point is located on the geographic scene unit layer S2 is calculated, and the precise position of the target sample point is assigned to the center point position of the scene unit with the greatest similarity.

2. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 1, characterized in that: Step (1) specifically includes: (1-1) Obtain scene description texts for all target samples and reference samples; (1-2) Recognizing the scene description text by using a preset regular expression to obtain the location description text and geographic scene feature values ​​of the target sample point, as well as the precise location and geographic scene feature values ​​of the reference sample point; (1-3) The location description text of the target sample point is processed according to the administrative address format to obtain the administrative location of the target sample point, and rough positioning is performed based on the administrative location to obtain the rough positioning location.

3. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 1, characterized in that: Step (2) specifically includes: (2-1) Generate target sample point distribution layer P1 based on the coarse positioning positions of all target sample points; (2-2) Generate a reference sample point distribution layer P2 based on the precise locations of all reference sample points.

4. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 1, characterized in that: Step (4) specifically includes: (4-1) According to the reference sample point distribution layer P2, obtain the scene unit corresponding to the reference sample point on the geographic scene unit layer S1; (4-2) For the numerical type geographic scene feature value of the reference sample point, using a membership function that is suitable for the change characteristics of the corresponding geographic scene feature, calculate the similarity between the geographic scene feature value and the corresponding geographic scene feature value of the scene unit corresponding to the reference sample point as the membership value of the geographic scene feature value; (4-3) for the enumerated type geographic scene feature value of the reference sample point, obtaining, according to a preset mapping rule, a similarity between the geographic scene feature value and the corresponding geographic scene feature value of the scene unit corresponding to the reference sample point, as the membership value of the geographic scene feature value, wherein the preset mapping rule is used to indicate the mapping relationship between the geographic scene feature value and the similarity; (4-4) According to the membership values ​​of all geographic scene feature values ​​of the reference sample points, the feature importance weight of each geographic scene feature is calculated using the entropy method, wherein the lower the membership value of the geographic scene feature, the greater the feature importance weight.

5. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 4, characterized in that: Step (4-2) specifically includes: (4-2-1) For geographic scene features of numerical type whose data variation does not exceed the preset range, a triangular membership function or a trapezoidal membership function is used to calculate the similarity between the corresponding geographic scene feature values ​​as the membership value of the corresponding geographic scene feature value; (4-2-2) For geographic scene features of numerical type whose data variation exceeds the preset range, any function among the polynomial membership function, Gaussian membership function, and root membership function is used to calculate the similarity between the corresponding geographic scene feature values ​​as the membership value of the corresponding geographic scene feature value.

6. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 5, characterized in that: The specific formula for calculating similarity using the triangular membership function is: μ=1-|xp-xs| / (xs-a) Where μ is the similarity, xp is the geographic scene feature value of the reference sample point, xs is the geographic scene feature value of the scene unit corresponding to the reference sample point on the geographic scene unit layer S1, and a is a variable parameter; The specific formula for calculating similarity using the trapezoidal membership function is: Where x′=xs±b, b is the neighborhood radius of the trapezoidal maximum value, and c is a variable parameter; The specific formula for calculating similarity using the Gaussian membership function is: Where σ is the standard deviation of the membership function; The specific formula for calculating similarity using the polynomial membership function is: μ=a0+a1|xp-xs|+a2x 2 |xp-xs| 2 +...+at|xp-xs| t Where a0, a1, ..., at are the coefficients of the polynomial; The specific formula for calculating similarity using the root membership function is: Where d is a variable parameter.

7. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 1, characterized in that: Step (5) specifically includes: (5-1) Filter out the township-level administrative units where each soil survey sampling point is located to generate the township layer C; (5-2) On the target sampling point distribution layer P1, a number of circular adjacent areas are divided with each soil survey sampling point as the center and a preset distance as the radius. The areas outside the circular adjacent areas on the target sampling point distribution layer P1 are deleted to generate a buffer layer B; (5-3) Perform an intersection operation on the geographic scene unit layer S1, the township layer C, and the buffer layer B to generate the geographic scene unit layer S2.

8. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 1, characterized in that: Step (6) specifically includes: (6-1) Solve the topological relationship between the target sample point distribution layer P1 and the geographic scene unit layer S2 to obtain the corresponding scene unit of each target sample point on the geographic scene unit layer S2; (6-2) extracting all geographic scene feature values ​​of the corresponding scene unit of each target sample point on the geographic scene unit layer S2; (6-3) By solving the topological relationship, all scene units adjacent to the scene unit corresponding to each target sample point are obtained, and all geographic scene feature values ​​of these adjacent scene units are extracted.

9. The method for automatically refining the fuzzy positions of historical soil survey sampling points according to claim 4, characterized in that: Step (7) specifically includes: (7-1) For different geographic scene feature values, select the membership function consistent with step (4), calculate the similarity between each geographic scene feature value of each target sample point and the corresponding geographic scene feature value of the corresponding scene unit on the geographic scene unit layer S2, and use it as the membership value of the geographic scene feature corresponding to the target sample point; (7-2) Taking the feature importance weight of the geographic scene feature as the corresponding weight, calculate the weighted sum of the membership value of each geographic scene feature of the target sample point as the similarity sim0 between each target sample point and the corresponding scene unit.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the computer program to implement the method according to any one of claims 1 to 9.

Images