A method for generating central collection points of spectral samples of homogeneous objects based on land type patches

Abstract

The present invention relates to a method for generating central acquisition points of homogeneous geological object spectral samples based on land type spots, and belongs to the field of remote sensing technology. First, the surface cover type spot data of the task area is obtained, and the constraint rules such as the spot area, grid size, spatial distribution, and acquisition range required for the generation of spectral acquisition points are set according to the task requirements, and the spectral acquisition point generation task is issued; secondly, according to the generated task, the geometric processing of the spots is completed, the acquisition standard grid is generated, the acquisition points are searched, and the acquisition point positions are adjusted; finally, the attribute information of the acquisition points is extracted. The method realizes the automatic generation of massive remote sensing image spectral acquisition points that meet custom rules based on spots, solves the problems of unreasonable spatial distribution of acquisition points and incorrect ground object attributes that occur in the processing of irregular spots such as narrow and long spots and central hole spots, and provides important basic support for the construction of spectral acquisition and database based on remote sensing images.

Description

Technical Field

[0001] The invention relates to the field of remote sensing technology, and in particular to a method for generating central acquisition points of homogeneous geological object spectrum samples based on land type patches. Background Art

[0002] In recent years, with the rapid development of remote sensing technology, the construction of a spectral database of typical land objects with a large area, long time series and complete categories based on remote sensing images, especially high-spectral satellite images, has become an important way of spectral data collection and database construction. It provides an economical, convenient and effective method to enrich the spectral data of different land objects under different states / background conditions.

[0003] The collection of spectral data is a very important part in the construction of the surface feature spectral database, and the generation and selection of spectral collection points are the basis of spectral data collection. Spectral collection points should be arranged in "homogeneous" areas with relatively single surface cover types, and have accurate surface feature target attribute information, reasonable spatial distribution, and can match the spatial resolution of remote sensing images used to collect spectral data. Generally, the surface patch data such as the surface cover type classification and survey results with high industry recognition published by national institutions can be used as the data source, and the patch data is obtained by combining the patch data with high spatial resolution remote sensing image data. For example, the national land survey data is an important achievement in clarifying the national land use status, covering cultivated land, gardens, forests, grasslands, wetlands, urban residential and industrial and mining land, transportation land, water areas and water conservancy facilities land and other surface cover types. Its patch data can provide accurate point information required for image pixel spectrum collection.

[0004] Usually, the method of generating points based on surface patch data is to calculate the geometric center point of the surface patch. First, the geometric figures of each patch are repaired, and then the patches that meet the task purpose are screened. The geometric center points of all patches that meet the requirements are calculated in turn, and the information such as surface cover type and administrative divisions are collected. Finally, the generated points are analyzed and summarized. There are the following defects in generating points using this method: in the process of point generation, it is difficult to strictly constrain the spatial distribution of points for narrow patches such as roads and rivers, irregularly shaped plots and central hole patches. At the same time, there is a situation where the geometric center point of the patch is located outside the patch. The generated point map cannot fully meet the requirement that the type of the collected point feature must be a "homogeneous" feature, and the matching degree with the spatial resolution of the image data used to collect the spectrum is poor, which is not conducive to the subsequent spectral data collection.

[0005] In response to the above problems, how to generate the center collection points of homogeneous object spectral samples based on the pattern in a more automated and intelligent manner and meet the requirements of remote sensing image spectral collection tasks, while ensuring the accuracy and reliability of the points, especially matching the spatial resolution scale of the image data used to collect spectral data, and ensuring that the type of ground object corresponding to the point is a "homogeneous" ground object, is an urgent problem to be solved in spectral collection work, and also determines the quality of remote sensing image pixel spectral collection data. Summary of the invention

[0006] The present invention aims at the problem of generating spectral acquisition points of image pixels and provides a method for generating central acquisition points of spectral samples of homogeneous objects based on land type patches. The method not only realizes the generation of large quantities of spectral acquisition point vector data, but also has the ability to define customized constraint rules for the generation of spectral acquisition points, which can improve the accuracy of generated acquisition points and the flexibility and editability of generation schemes, and facilitates the large-scale intelligent generation of central acquisition points of spectral samples of homogeneous objects in large areas and large ranges.

[0007] In order to achieve the above object, the technical solution provided by the present invention is a method for generating center acquisition points of a homogeneous object spectrum sample based on a planar pattern, comprising the following steps:

[0008] Step 1: Configure the point generation plan. Generate a plan template based on the spectrum acquisition point layout principles and requirements, and combine the work area homogeneous object spectrum sample collection requirements to generate the required map area, spatial distribution, collection range, grid size and other task requirements to form a point generation plan that meets the spectrum sample collection requirements;

[0009] Step 2: Processing of patch geometry. First, check the data validity of the patches in the specified area, repair the wrong / invalid geometry, calculate the area of ​​the repaired patches, and select the basic patch data for subsequent processing according to the patch area constraints described in step 1;

[0010] Step 3, generate a standard acquisition grid. According to the grid size constraint described in step 1, that is, combined with the spatial resolution scale of the remote sensing image to be used for spectral data acquisition, generate a standard acquisition grid, where the grid size is 30m×30m by default; rasterize the basic image spots described in step 2 based on the standard grid to generate image spot grids, and obtain the surface coverage type corresponding to the standard grid;

[0011] Step 4, search for collection points. According to the collection range constraints described in step 1, ensure that the spatial range centered on the ground feature spectral sample collection point belongs to the same surface coverage type, that is, meets the requirements of "homogeneous" ground features. Assign different weights to each standard grid, and ensure the spatial distribution of the ground feature spectral sample collection points according to the spatial distribution constraints described in step 1. Use the setting of the search dirty bit mode to determine whether the current search is repeated. If not, continue to generate collection points to obtain the initial collection points;

[0012] Step 5: Adjust the acquisition points, perform biquadratic linear interpolation based on the characteristics of the spots and the initial distribution of the acquisition points, adjust the acquisition points and generate acquisition points according to the grid, so that the acquisition points in the spots are scattered in the internal area of ​​the spots, meet the number and distribution requirements of the points generated by the large and small types of spots of the surface cover type, and obtain the final distribution of the acquisition points;

[0013] Step 6: extract the collection point information, automatically obtain the surface cover type, administrative division and other information based on the acquired collection points, and uniformly manage the final collection point data.

[0014] Among them, the configuration of the ground object spectral sample collection plan described in step 1 is to set the constraint rules such as the map area, grid size, spatial distribution, and collection range required for the generation of actual ground object spectral sample collection points in combination with task requirements, and complete the configuration of the collection point generation plan to facilitate subsequent collection point generation based on the constraint rules.

[0015] The geometric processing of the spots described in step 2 should meet the normative requirements of shapefiles and feature classes stored in personal geographic databases or file geographic databases, and perform data validity checks. For invalid geometric figures in the spots, geometric problems are repaired, and the area of ​​the geometrically repaired spots is counted to screen out the spots that meet the spot area constraints in the point generation plan.

[0016] The data validity check mainly includes the following geometric problems: 1. Short line segment - the line segment is shorter than the length allowed by the system unit referenced by the control associated with the geometry; 2. Empty geometry - the element does not have a geometric rule or the SHAPE attribute field is empty; 3. Incorrect ring direction - the ring of the face is not correctly oriented (outer ring - clockwise, inner ring - counterclockwise); 4. Incorrect line segment direction - the orientation of each line segment is inconsistent, such as the "end" point of line segment i is not connected to the "start" point of line segment i+1; 5. Self-intersection - the edge of the polygon should not intersect with its other edges; 6. Non-closed loop - the "end" point of the last line segment in the loop must be connected to the "start" point of the first line segment; 7. Empty component - the geometry has multiple components, one of which is empty (no geometry); 8. Duplicate vertices - the coordinates of two or more vertices of the geometry overlap; 9. Unmatched attributes - the Z coordinate or M coordinate of a segment endpoint does not match the Z coordinate or M coordinate of the corresponding endpoint in the next segment; 10. Discontinuous part - a part of the geometry consists of disconnected or discontinuous parts; 11. Empty Z value - one or more vertices of the geometry have empty Z values ​​(for example, NaN).

[0017] The invalid geometric figures are repaired according to geometric figure rules.

[0018] The geometrically restored patches are used to calculate the area of ​​each patch, and according to the patch area constraints of the point generation scheme in step 1, patches that meet the task requirements are screened out as basic patch data.

[0019] Among them, the method for generating a standard acquisition grid described in step 3 is based on the grid size constraint described in step 1, that is, the acquisition standard grid is generated with reference to the spatial resolution size scale of the remote sensing image for collecting spectral sample data of the object to be collected. The rasterization of the spot vector data based on the standard grid described in step 2, that is, based on the spatial parameters of the standard grid, the vector data of the spot data is converted into raster data. The spot raster is consistent with the standard grid in geographic space, and the center coordinates of the standard grid are used to query the spot raster to obtain the surface coverage type corresponding to the standard grid.

[0020] Among them, the method for searching and collecting points described in step 4 searches the standard grids one by one in the eight-neighborhood search mode, ensures that the spatial range centered on the ground object spectrum sample collection point belongs to the same surface coverage type according to the collection range constraint described in step 1, assigns different weights to each standard grid, and ensures that the spatial distribution of the collection points meets the collection requirements according to the spatial distribution constraint described in step 1. By setting the search dirty bit, determine whether the current search is repeated, and if not, continue to generate the collection points to obtain the initial collection points.

[0021] Among them, the adjustment of the collection point position described in step 5 uses the bilinear quadratic interpolation method to adjust the position of the initial ground feature spectral sample collection points described in step 4, and the collection position is adjusted based on the standard grid described in step 3, so that the distance between each collection point and the upper, lower, left and right boundaries of the standard grid of the same surface cover type is consistent, and the quantity and distribution requirements of the points generated by the large and small types of surface cover types are met, so as to obtain the final collection points.

[0022] Among them, the collection point information is extracted in step 6, and the collection points in step 5 will inherit the surface coverage type of the standard grid in step 3. After obtaining the surface coverage type, administrative division and other information, the final collection points are uniformly managed in shapefile vector data format.

[0023] The present invention provides a method for generating central collection points of homogeneous ground object spectrum samples based on ground type patches, and the method has the advantages and effects of: the method can realize the generation of a large number of ground object spectrum sample collection points that meet spatial constraints based on planar patches, solves the problems of unreasonable spatial distribution of ground object spectrum sample collection points and incorrect surface coverage types that occur during the processing of irregular ground type patches such as narrow and long patches such as roads and rivers, 7-shaped and V-shaped geometric center points outside the patches, and central hole patches, based on operations such as patch area screening and patch vector data rasterization processing, improves the efficiency of spatial operations, logical operations, and attribute connections, has the ability to generate ground object spectrum sample collection vector point data based on custom rules such as patch area and spatial distribution, can realize rapid, batch, automatic and intelligent generation of collection points that meet the needs of spectrum collection tasks, and is convenient for obtaining point data that meet the needs of subsequent spectrum collection tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The following is a business process diagram of an embodiment of the present invention.

[0025] Figure 2 Schematic diagram of geometric restoration of a patch according to an embodiment of the present invention.

[0026] Figure 3 Schematic diagram of converting vector data into raster data according to an embodiment of the present invention.

[0027] Figure 4 The figure is a schematic diagram of the collection point generation process according to an embodiment of the present invention.

[0028] Figure 5 This is a schematic diagram of generating points for collecting ground features with holes according to an embodiment of the present invention.

[0029] Figure 6 A schematic diagram of generating central acquisition points of large-scale ground object spectral samples according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The present invention provides a method for generating central collection points of homogeneous geological spectra samples based on land type spots. The method can realize the generation of a large number of homogeneous geological spectra sample collection points that meet spatial constraints based on spots, solves the problems of unreasonable spatial distribution of collection points and incorrect surface coverage types that occur during the processing of irregular spots such as narrow spots such as roads and rivers, 7-shaped and V-shaped geometric center points outside the spots, and central hole spots, etc., can realize batch and automatic generation of collection points that meet the needs of spectral collection tasks, and has the ability to generate spectral collection vector point data based on custom rules such as spot area and spatial distribution. Taking the generation of homogeneous geological spectra sample collection points based on spots in Pingyuan County, Dezhou City, Shandong Province as an example, the technical solution of the present invention is further explained in combination with the accompanying drawings and embodiments.

[0031] like Figure 1 As shown, the process of the embodiment of the present invention includes the following steps:

[0032] Step 1: Obtain data from the mission area of ​​Pingyuan County, Dezhou City, Shandong Province. According to the mission requirements, set the constraint rules such as the area of ​​the actual spectrum collection points, grid size, spatial distribution, and collection range required for the generation of the actual spectrum collection points, and form a plan for the generation of the collection points, which specifically includes the following two steps:

[0033] Step 1.1, obtain the regional range vector file and regional land cover type patch data of the mission area of ​​Pingyuan County, Dezhou City, Shandong Province;

[0034] Step 1.2: According to the task requirements, it is necessary to sample the hyperspectral image data with a spatial resolution of 30m×30m. According to this requirement, set the constraint rules such as the area of ​​the patch, grid size, spatial distribution, and acquisition range required by the actual acquisition point generation. The patch area is set to be larger than 3×3 hyperspectral image pixels, which is 10000m 2 The grid size is consistent with the spatial resolution of the hyperspectral image data, that is, it is set to 30m×30m; the spatial distribution specifies the position and distribution of the points in the grid, the position is set to be centered, and the sampling point distribution is set to 5km, that is, the interval between points of the same surface cover type is 5km; the acquisition range specifies the spatial sampling range and attribute sampling range, and is set to sample the entire range and the spatial range and attribute range of all surface cover types.

[0035] Step 2: According to the acquisition point generation plan, complete the geometric processing of the pattern, generate the acquisition standard grid, search for the acquisition points and adjust the acquisition point positions, which specifically includes the following two steps:

[0036] Step 2.1, check the validity of the patch data within the task area, and check the polygons in the patch data for compliance with the normative requirements, that is, conduct a comprehensive check on the length of line segments, empty geometry, empty components, the direction of rings and line segments, the intersection of edges, the closure of rings, repeated vertices, mismatched attributes, discontinuous parts, and empty Z values; for polygons that do not meet the norms, the geometry needs to be repaired, and for non-standard geometry, it is repaired according to the normative rules, such as Figure 2 As shown, the error is that there is overlap between features. The geometry repair method is to perform geometry repair according to the overlapping part of an adjacent polygon and the rule is to perform geometry repair according to the shortest common edge.

[0037] Step 2.2, calculate the area of ​​the geometrically restored patch. The patch area uses the ellipse area. The projection coordinates are calculated by coordinate inversion to obtain the spherical longitude and latitude of each patch point. The ellipsoid area is calculated using the area calculation model of any trapezoidal block on the ellipsoid surface, thereby obtaining the ellipsoid area of ​​the patch. The constant values ​​such as the ellipsoid parameters used in the calculation are the same as those used in the theoretical area calculation of the map sheet. Preliminary screening of patches that meet the task requirements is performed based on the patch area constraint. The patch area is compared with the patch area constraint. Patches that are smaller than the area constraint (10000m 2 ) spots are removed and do not participate in subsequent calculations, which can improve the efficiency of collecting point generation.

[0038] Step 3: Based on the grid size constraint and the spatial resolution of the remote sensing image data to be used for spectral data acquisition, a 30m×30m acquisition standard grid is generated. Based on the spatial parameters of the standard grid, the vector data of the spot data is converted into raster data, where the raster value corresponds to the spot identification number, such as Figure 3 As shown, the positions of raster and vector data are judged based on spatial information. When the raster is completely inside the vector data, the complete raster is retained. When the raster is at the boundary of the vector data, the raster is cropped based on the vector boundary and the raster inside the vector is retained. The surface coverage type of the raster is consistent with the surface coverage type of the vector.

[0039] Step 4: Search for collection points and obtain the initial collection points by using eight-neighborhood search and setting search dirty positions. Specifically, it includes the following two steps:

[0040] Step 4.1: Search the generated standard grids one by one in the eight-neighborhood search method. According to the acquisition range constraint, ensure that the spatial range centered on the acquisition point belongs to the same surface coverage type: starting from the current point, count the number of whole layers of grids with the same surface coverage type as the current pixel. If a fault is encountered, the calculation is interrupted. If the surface coverage type of the grids with a maximum range of 3×3 pixels around it is the same as the current point, the weight of the current point is set to 9. Traverse the points according to this method and assign different weights to each standard grid, such as Figure 4 As shown;

[0041] Step 4.2, according to the spatial distribution constraints set in step 1.2, ensure the spatial distribution of the collection points, that is, the position is set to the center, and the interval between points of the same surface cover type is 5km. For the searched points, set their search flag to 1, and determine whether the current search is repeated. If not, continue to generate collection points to obtain the initial collection points.

[0042] Step 5, use the bilinear quadratic interpolation method to adjust the position of the initial collection points: Based on the weight value of the grid participating in the weight calculation in step 4.1, the collection points are adjusted in the x and y directions of the current grid using the bilinear quadratic interpolation method to keep the distance between each collection point and the upper, lower, left and right boundaries of the current grid consistent. Finally, the collection point generation results of the features with holes and large-scale features are as follows: Figure 5 , Figure 6 shown.

[0043] Step 6: Extract the collection point information. This includes the following two steps:

[0044] Step 6.1, based on the relative spatial position of the collection point and the standard grid, find the grid value corresponding to the collection point. Since the grid value is equal to the spot identification number, the surface cover type attribute information, administrative division, point latitude and longitude and other information of the grid can be obtained from the spot attribute information, and it is used as the surface cover type attribute information of the sampling point;

[0045] Step 6.2, save the collected points output as a shapefile format file. The current collection point generation task is completed, and 18,347 homogeneous ground object spectral sample collection points that meet the spatial constraints such as accurate coverage of corresponding ground object type attributes and reasonable spatial distribution in Pingyuan County, Dezhou City, Shandong Province are obtained, including 13,535 vegetation features, 487 water features, 9 soil features, and 4,266 artificial target features. This effectively solves the problem of generating collection points and the "homogeneity" requirement of the corresponding ground objects in the work of collecting homogeneous ground object target pixel spectral data based on massive hyperspectral image data, and provides the most important guarantee for the authenticity and correctness of basic point information for the effective collection and iterative update of batch and intelligent pixel spectral data.

[0046] In addition to the above embodiments, the present invention may also have other implementation modes. Any technical solution formed by equivalent replacement or equivalent transformation falls within the protection scope required by the present invention.

Claims

1. A method for generating central collection points of homogeneous geological spectral samples based on land type patches, characterized in that: The steps include: Step 1, configure the point generation plan; generate a plan template based on the spectrum acquisition point layout principles and requirements, and generate the required constraint rules of the spot area, spatial distribution, acquisition range, and grid size in combination with the spectrum sample acquisition requirements of homogeneous objects in the work area to form a point generation plan that meets the spectrum sample acquisition requirements; Step 2, processing of the geometric figures of the spots; firstly, checking the data validity of the spots in the specified area, repairing the invalid geometric figures, calculating the areas of the repaired spots in turn, and selecting the basic spot data for the subsequent processing process according to the spot area constraints described in step 1; Step 3, generating a standard acquisition grid; according to the grid size constraint described in step 1, that is, combining the spatial resolution scale of the remote sensing image to be used for spectral data acquisition, generating a standard acquisition grid, wherein the grid size is 30m×30m by default; rasterizing the basic spots described in step 2 based on the standard grid, generating a spot raster, and obtaining the surface coverage type corresponding to the standard grid; Step 4, search for collection points; search the standard grids one by one in the eight-neighborhood search method; according to the collection range constraint rules described in step 1, ensure that the spatial range centered on the ground feature spectral sample collection point belongs to the same surface coverage type, that is, meets the requirements of homogeneous ground features; assign different weights to each standard grid, and ensure the spatial distribution of the ground feature spectral sample collection points according to the spatial distribution constraints described in step 1; use the setting search dirty bit mode to determine whether the current search is repeated, if not repeated, continue to generate collection points to obtain initial collection points; Step 5: Adjust the acquisition points, perform biquadratic linear interpolation based on the characteristics of the spots and the initial distribution of the acquisition points, adjust the acquisition points and generate acquisition points according to the grid, so that the acquisition points in the spots are scattered in the internal area of ​​the spots, meet the number and distribution requirements of the points generated by the large and small types of spots of the surface cover type, and obtain the final distribution of the acquisition points; Step 6: extract the collection point information, automatically obtain the surface cover type and administrative division information based on the acquired collection points, and uniformly manage the final collection point data.

2. The method for generating central collection points of homogeneous geological spectral samples based on land type patches according to claim 1, characterized in that: The validity check of the image spot data includes: (1) Short segment - a segment that is shorter than the length allowed by the system unit referenced by the control associated with the geometry; (2) Empty geometry - the feature has no geometry or the SHAPE attribute field is empty; (3) Incorrect ring orientation - the rings of the surface are not oriented correctly, where the outer ring is clockwise and the inner ring is counterclockwise; (4) Incorrect line segment orientation - the orientation of each line segment is inconsistent. The "end" point of line segment i should be connected to the "start" point of line segment i+1; (5) Self-intersection - the edges of a polygon should not intersect with its other edges; (6) Non-closed loop - the "end" point of the last line segment in the loop must be connected to the "start" point of the first line segment; (7) Empty parts - a geometry has multiple parts, one of which is empty; (8) Duplicate vertices - the coordinates of two or more vertices of the geometry overlap; (9) Mismatched attributes - the Z coordinate or M coordinate of a line segment endpoint does not match the Z coordinate or M coordinate of the corresponding endpoint in the next line segment; (10) Discontinuous part - a part of geometry consists of disconnected or discontinuous parts; (11) Empty Z Values ​​- One or more vertices of the geometry have empty Z values.

3. The method for generating central collection points of homogeneous geological spectral samples based on land type patches according to claim 1, characterized in that: Geometry repairs include: Filter out invalid geometries; Repairs are performed based on the geometry check type to ensure that they meet the specifications of the patch elements.

4. The method for generating central collection points of homogeneous geological spectral samples based on land type patches according to claim 1, characterized in that: The patch area calculation includes: Calculate the projection area based on the geographic coordinate information of the map data.

5. The method for generating central collection points of homogeneous geological spectral samples based on land type patches according to claim 1, characterized in that: The rasterization processing of the spot vector data to obtain the surface coverage type corresponding to the standard grid includes: Based on the spatial parameters of the standard grid, the vector data of the spot data is converted into raster data; The center coordinates of the standard grid are used to query the patch grid to obtain the surface cover type.

Images