A method, device, electronic device and storage medium for generating a pattern map based on spatial information

Abstract

The present invention relates to a method for generating a spot map based on spatial information, comprising: obtaining super-resolution remote sensing image data of different phases; performing spot change detection and spot semantic segmentation recognition on the pre-processed super-resolution remote sensing image data, and combining the detection result and the recognition result to obtain a spot change information set; traversing and screening out adjacent spots that meet the merging conditions in the spot change information set to obtain a spot information set to be merged; performing slope filtering on the spot set to be merged according to elevation data to obtain a slope filtered spot set, and performing consistency check on the slope filtered spot set to obtain a spot set of elements of the same nature; merging the slope filtered spot set according to the spots of the element spot set of the same nature to obtain map data of completely segmented spots. The method for generating a spot map based on spatial information of the present invention is not affected by environmental factors such as temperature, humidity and precipitation, and the spot merging result has high reliability and stability.

Description

Technical Field

[0001] The present invention relates to the field of super-resolution map drawing, and in particular to a method, device, electronic equipment and computer storage medium for generating a spot map based on spatial information. Background Art

[0002] The process of traditional map drawing, map spots or plot planning usually relies on GIS (Geographic Information System), and it is necessary to manually mark the undivided areas in the map to draw the map spots or plots. Among them, the map spots are to classify the plots with basically the same landforms, land use types, soil erosion, etc. into one category, and draw them on the map to form a map spot. The plot is a unit of a terrain type.

[0003] Nowadays, with the popularity of remote sensing images, neural networks are usually used to divide, draw, and annotate areas in remote sensing images to achieve the drawing of map spots, plots, etc. However, due to the characteristics of remote sensing images such as super-resolution and image temporal changes, if neural networks are used to draw spots or plots, the remote sensing images need to be sliced ​​before the sliced ​​remote sensing images can be input into the neural network for instance segmentation, semantic segmentation, recognition, etc. Therefore, sliced ​​remote sensing images are prone to fragmentation (i.e., one spot is divided into multiple adjacent spots) and duplication (i.e., one spot is covered by another spot) in the segmentation and recognition results of the neural network. In particular, remote sensing images also have multi-temporal characteristics. For remote sensing images of the same area at different times, the features extracted by the neural network are prone to change, further affecting the accuracy of the segmentation and recognition of remote sensing images by the neural network. Therefore, it is necessary to further filter and screen the output results of the neural network.

[0004] In order to solve the problems of fragmentation and duplication when segmenting remote sensing images, existing technologies generally analyze the spectral information of remote sensing images and exclude abnormal results based on the analysis results to perform operations such as merging spots or plots. However, since spectral information is easily affected by factors such as temperature, humidity, and precipitation, it is easy to cause analysis errors, which ultimately leads to inaccurate results of merging spots and plots. Summary of the invention

[0005] Based on this, the object of the present invention is to provide a method for generating a patch map based on spatial information.

[0006] A method for generating a spot map based on spatial information comprises the following steps:

[0007] S1: Acquire super-resolution remote sensing image data at different phases;

[0008] S2: Performing spot change detection and spot semantic segmentation recognition on the preprocessed super-resolution remote sensing image data, and combining the detection results with the recognition results to obtain a spot change information set;

[0009] S3: traverse and filter out the adjacent spots that meet the merging conditions in the spot change information set, and obtain the spot information set to be merged;

[0010] S4: performing slope filtering on the patch set to be merged according to elevation data to obtain a slope filtered patch set, and performing consistency check on the slope filtered patch set to obtain a patch set of elements of the same nature;

[0011] S5: Merge the slope filtering patch set according to the patches of the feature patch set of the same nature to obtain map data of the complete segmented patches.

[0012] The method for generating a patch map based on spatial information described in the present invention combines the results of patch change detection and patch semantic segmentation to analyze the degree of change of patches in the same area in different time phases, thereby reducing the segmentation error caused by the multi-temporal characteristics of remote sensing images. In addition, the present invention performs slope analysis through elevation data to analyze whether the slope continuity of the boundary meets the merged slope, thereby ensuring the accuracy of patch merging. At the same time, consistency checks are performed to ensure that the properties of the plots between the merged patches are consistent, and further consider the consistency of the merged patches in land use types or other business data. In addition, the present invention screens, filters, and merges patches by combining spatial information with non-spatial information, and is not affected by environmental factors such as temperature, humidity, and precipitation, thereby improving the reliability and stability of patch merging results.

[0013] Furthermore, the step S2 comprises the following steps:

[0014] S21: Slice and match the super-resolution remote sensing image data to obtain a number of tile pairs at different times and locations;

[0015] S22A: performing convolution and contrast detection on a plurality of tile pairs in sequence, and calculating the area of ​​the detection results to obtain the change areas of the plurality of tile pairs;

[0016] S22B: performing convolution feature extraction and semantic segmentation on a number of tile pairs to obtain semantic information of the number of tile pairs;

[0017] S23: combining the changed regions of all tile pairs with the corresponding semantic information to obtain a patch change information set;

[0018] Among them, the patch change information set includes the longitude and latitude boundary point set of the changed area, the area of ​​the changed area and the semantic change status of the changed area; the longitude and latitude boundary point set of the changed area is the longitude and latitude boundary point set P of the patch, the area of ​​the changed area is the area A of the changed patch corresponding to the longitude and latitude boundary point set P, and the semantic status of the changed area is the semantic information C of the tile pair corresponding to the area A of the changed patch.

[0019] The present invention uses tile pairs to ensure that the image spots at multiple time points are accurately compared for the same geographical location, thereby ensuring the consistency of multi-temporal data processing; then, the changes in the time series in the tile pairs are detected through changes to increase the basis for merging decisions.

[0020] Furthermore, step S3 includes the following steps:

[0021] S31: Process all spots in the spot change information set to obtain a spot information set;

[0022] S32: performing spatial segmentation calculation on the image spot information set to obtain a Voronoi diagram of the image spot information set;

[0023] The spatial segmentation calculation is specifically expressed as follows:

[0024] The Delaunay triangulation network is formed by the spots in the spot information set, and the spots and triangles are numbered to divide the spot combinations that form the triangles;

[0025] Calculate the circumcenters of all triangles according to the triangle spot combination to obtain the circumcenter of the triangle, then connect the circumcenters of adjacent triangles with common edges and save them in a one-dimensional Novo graph linked list;

[0026] The perpendicular bisector rays corresponding to the sides of the triangles that do not have common sides are saved in the Voronoi diagram linked list. After the traversal is completed, the Voronoi diagram of the graph spot information set is obtained;

[0027] S33: Traverse and filter the spot information set according to the adjacency relationship of the Voronoi graph to obtain the spot set to be merged.

[0028] Compared with the prior art that directly queries adjacent objects through patches, the present invention uses Voronoi diagrams to represent the adjacency relationship between patches and reduces the time complexity from O(n 2 ) is optimized to O(nlogn), making the subsequent spatial analysis more efficient.

[0029] Furthermore, the step S4 comprises the following steps:

[0030] S41: performing slope analysis filtering on the patch set to be merged through elevation data to obtain a slope filtering patch set;

[0031] The analysis and filtering is to perform quadratic surface fitting on the landforms between the spots in the set of spots to be merged to obtain landform continuity, and perform interpolation analysis based on the local landform continuity to obtain the local landform elevation difference, retain the spots whose local landform elevation difference is lower than a reference factor, and delete the spots whose local landform elevation difference exceeds a reference factor from the set of spots to be merged to obtain a slope filtered spot set;

[0032] S42: performing consistency check on the slope filtering patch set to obtain a patch set of elements of the same nature;

[0033] The consistency check is to determine whether the plots to be merged in the slope filtering plot set belong to the same type of plots in nature based on land survey data, and store the plots that belong to the same type in a plot set of elements with the same nature.

[0034] The present invention analyzes the topographic continuity and elevation changes between map patches through elevation data to ensure that the merged map patches are reasonable in terms of terrain. In addition, through the information in the land survey data, the merged map patches are considered to be consistent in business logic, making the merged map patches more in line with policy making, resource management, etc.

[0035] A spot map generating device based on spatial information, comprising a remote sensing image data acquisition unit, a multi-temporal spot change information acquisition unit, a spot screening unit, a spot filtering unit and a spot merging unit;

[0036] The remote sensing image data acquisition unit is used to acquire super-resolution remote sensing image data of different phases;

[0037] The multi-temporal spot change information acquisition unit is used to perform spot change detection and spot semantic segmentation recognition on the pre-processed super-resolution remote sensing image data, and combine the detection result and the recognition result to obtain a spot change information set;

[0038] The spot screening unit is used to traverse and screen out adjacent spots that meet the merging conditions in the spot change information set to obtain the spot information set to be merged;

[0039] The patch filtering unit is used to perform slope filtering on the patch set to be merged according to elevation data to obtain a slope filtered patch set, and to perform consistency check on the slope filtered patch set to obtain a patch set of elements of the same nature;

[0040] The patch merging unit is used to merge the patches in the slope filtering patch set that conform to the element patch set of the same nature, so as to obtain the map data of the complete segmented patches.

[0041] Furthermore, the multi-temporal spot change information acquisition unit includes a remote sensing image slice matching module, a change region detection module, a semantic segmentation neural network and a spot change information combination module;

[0042] The remote sensing image slice matching module is used to slice and match the super-resolution remote sensing image data to obtain a number of tile pairs that are co-located at different times;

[0043] The change region detection module is used to perform convolution and contrast detection on a plurality of tile pairs in sequence, and calculate the region of the detection result to obtain the change region of the plurality of tile pairs;

[0044] The semantic segmentation neural network is used to perform convolution feature extraction and semantic segmentation on a plurality of tile pairs to obtain semantic information of the plurality of tile pairs;

[0045] The patch change information combining module is used to combine the change areas of all tile pairs with the corresponding semantic information to obtain a patch change information set;

[0046] Among them, the patch change information set includes the longitude and latitude boundary point set of the changed area, the area of ​​the changed area and the semantic change status of the changed area; the longitude and latitude boundary point set of the changed area is the longitude and latitude boundary point set P of the patch, the area of ​​the changed area is the area A of the changed patch corresponding to the longitude and latitude boundary point set P, and the semantic status of the changed area is the semantic information C of the tile pair corresponding to the area A of the changed patch.

[0047] Furthermore, the spot screening unit includes a spot centroid acquisition module, a spot Voronoi diagram conversion module and a spot Voronoi diagram screening module;

[0048] The spot centroid acquisition module is used to process all spots in the spot change information set to obtain the spot information set;

[0049] The image spot Voronoi diagram conversion module is used to perform spatial segmentation calculation on the image spot information set to obtain the Voronoi diagram of the image spot information set;

[0050] The spatial segmentation calculation is specifically expressed as follows:

[0051] The Delaunay triangulation network is formed by the spots in the spot information set, and the spots and triangles are numbered to divide the spot combinations that form the triangles;

[0052] Calculate the circumcenters of all triangles according to the triangle spot combination to obtain the circumcenter of the triangle, then connect the circumcenters of adjacent triangles with common edges and save them in the Voronoi graph linked list;

[0053] The perpendicular bisector rays corresponding to the sides of the triangles that do not have common sides are saved in the Voronoi diagram linked list. After the traversal is completed, the Voronoi diagram of the graph spot information set is obtained;

[0054] The image spot Voronoi diagram screening module is used to traverse and screen the image spot information set according to the adjacency relationship of the Voronoi diagram to obtain the image spot set to be merged.

[0055] Further, the spot filtering unit includes a spot slope filtering module and a spot consistency filtering module;

[0056] The patch slope filtering module is used to perform slope analysis filtering on the patch set to be merged through elevation data to obtain a slope filtered patch set;

[0057] The analysis and filtering is to perform quadratic surface fitting on the landforms between the spots in the set of spots to be merged to obtain landform continuity, and perform interpolation analysis based on the local landform continuity to obtain the local landform elevation difference, retain the spots whose local landform elevation difference is lower than a reference factor, and delete the spots whose local landform elevation difference exceeds a reference factor from the set of spots to be merged to obtain a slope filtered spot set;

[0058] The patch consistency filtering module is used to perform consistency check on the slope filtering patch set to obtain a patch set of elements of the same nature;

[0059] The consistency check is to determine whether the plots to be merged in the slope filtering plot set belong to the same type of plots in nature based on land survey data, and store the plots that belong to the same type in a plot set of elements with the same nature.

[0060] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a schematic diagram of the structure of the device for generating a spot map based on spatial information according to the present invention;

[0062] Figure 2 It is a flow chart of the method for generating a spot map based on spatial information according to the present invention;

[0063] Figure 3 It is a simple schematic diagram of the area of ​​the patch corresponding to the centroid;

[0064] Figure 4 A specific schematic diagram of the space segmentation calculation of the present invention;

[0065] Figure 5 It is a simple schematic diagram of the present invention for performing image spot screening through the adjacency relationship of the Voronoi diagram;

[0066] Figure 6 This is a simple schematic diagram of the third national land survey data;

[0067] Figure 7 A simplified schematic diagram of a 3×3 moving window of the target pixel. DETAILED DESCRIPTION

[0068] In order to solve the problems of fragmentation, duplication and low accuracy of merging results in the process of using remote sensing images for image segmentation and merging, the present invention effectively screens, filters and merges the segmented images by combining spatial and non-spatial information such as adjacency relationships of image spots, business data, elevation data (Digital Elevation Model, DEM), thereby achieving a high confidence level in the merging of remote sensing images after image segmentation, so as to ensure the accuracy and reliability of the final generated map data in practical applications.

[0069] Based on the above design, the present invention proposes a method for generating a spot map based on spatial information, and based on this method, proposes a device for generating a spot map based on spatial information.

[0070] Please also see Figure 1 and Figure 2 , Figure 1 Schematic diagram of the structure of the device for generating a spot map based on spatial information according to the present invention. Figure 2 The present invention is a flowchart of the method for generating a patch map based on spatial information.

[0071] The spot map generating device based on spatial information comprises a remote sensing image data acquiring unit 1, a multi-temporal spot change information acquiring unit 2, a spot screening unit 3, a spot filtering unit 4 and a spot merging unit 5.

[0072] The remote sensing image data acquisition unit 1 is used to execute step S1: acquiring super-resolution remote sensing image data of different phases.

[0073] Specifically, rectangular areas with the same position at different time sequences in super-resolution remote sensing images are collected as super-resolution remote sensing image data.

[0074] In this embodiment, the remote sensing image is cropped by using GIS (Geographic Information System) processing software, that is, by inputting the latitude and longitude of the upper left corner and the latitude and longitude of the lower right corner, the corresponding rectangular area is marked in the remote sensing image to obtain a rectangular remote sensing image with consistent longitude and latitude of the rectangular diagonal. However, due to different cropping methods, the cropping methods used by users are different, and the present invention does not specifically limit and elaborate on them here.

[0075] The multi-temporal spot change information acquisition unit 2 is used to execute step S2: performing spot change detection and spot semantic segmentation recognition on the preprocessed super-resolution remote sensing image data, and combining the detection result and the recognition result to obtain a spot change information set.

[0076] Specifically, the multi-temporal spot change information acquisition unit 2 includes a remote sensing image slice matching module 21 , a change region detection module 22A, a semantic segmentation neural network 22B and a spot change information combination module 23 .

[0077] The remote sensing image slice matching module 21 is used to execute step S21: slicing and matching the super-resolution remote sensing image data to obtain a plurality of tile pairs that are co-located at different times.

[0078] Specifically, the super-resolution remote sensing image data is processed by padding, and the padded super-resolution remote sensing image data is sliced ​​with a size of 512×512 to obtain tiles, and the tiles of different phases and the same position are combined and matched according to the position information to obtain several tile pairs of different phases but the same position.

[0079] Among them, the tile is a tile of a tile map pyramid model. The tiles are stored and displayed at different resolutions according to user needs, forming a pyramid structure with resolutions from coarse to fine and data volumes from small to large. The represented geographical scope remains unchanged, but the lower the pyramid, the more detailed the map information represented and the larger the scale.

[0080] The change region detection module 22A is used to execute step S22A: perform convolution and contrast detection on a plurality of tile pairs in sequence, and calculate the area of ​​the detection results to obtain the change regions of the plurality of tile pairs.

[0081] Specifically, the change region detection module 22A includes a change detection neural network and an area calculation module; the change detection neural network is a neural network with a convolution structure, which is used to perform convolution feature extraction and comparison calculation on tile pairs at different phases and the same position to identify the two-dimensional geometric area of ​​the change spot in the tile pair and obtain the pixel boundary point set of the change spot; the area calculation module is an area calculation formula calculation or integral calculation function in the geographic information system (GIS), and the area calculation module is used to convert the pixel boundary point set into longitude and latitude to obtain the longitude and latitude boundary point set of the change spot, and calculate the area of ​​the change spot based on the longitude and latitude boundary point set to obtain the change region.

[0082] The comparison calculation uses Euclidean distance or Manhattan distance. The latitude and longitude boundary point set P consists of a series of points p i =(lat i ,lon i ) is used to represent the longitude and latitude boundaries of the change patch. i Represented as the current point p i Latitude, lon i Represented as the current point p i longitude.

[0083] The present invention does not specifically limit which convolutional neural network is used for feature extraction and comparison calculation.

[0084] The semantic segmentation neural network 22B is used to execute step S22B: perform convolution feature extraction and semantic segmentation on a plurality of tile pairs to obtain semantic information of the plurality of tile pairs.

[0085] Specifically, the semantic segmentation neural network is a network with similar structures such as U-Net (U-type network) and FCN (fully convolutional network), and its structure includes a feature extraction module and a semantic segmentation module; the feature extraction module performs convolution feature extraction on the input image data several times to obtain a feature map; the semantic segmentation module performs semantic analysis on the features of the feature map to obtain the semantic information of each tile in all tile pairs, that is, to obtain the semantic information of several tile pairs.

[0086] The semantic information C of the tile pair is specifically represented by the corresponding different phases (type before ,type after ), its type before Indicates the semantic information of the previous phase tile in the tile pair, type after Indicates the semantic information of the tile in the next phase after the tile is matched.

[0087] The patch change information combining module 23 is used to execute step S23: combining the change regions of all tile pairs with the corresponding semantic information to obtain a patch change information set.

[0088] Specifically, according to the set of longitude and latitude boundary points in the change area of ​​the tile pair, the semantic information of the corresponding tile pair is allocated to a set to obtain a patch change information set.

[0089] The patch change information set includes the latitude and longitude boundary point set of the change area, the area of ​​the change area, and the semantic change of the change area. The latitude and longitude boundary point set of the change area is the latitude and longitude boundary point set P of the patch, the area of ​​the change area is the area A of the change patch corresponding to the latitude and longitude boundary point set P, and the semantic situation of the change area is the semantic information C of the tile pair corresponding to the area A of the change patch.

[0090] Among them, by using the patch change information set to make the semantic information and area corresponding to the same tile consistent, the computer can filter, screen and merge the patches more efficiently in subsequent steps.

[0091] The spot screening unit 3 is used to execute step S3: traverse and screen out adjacent spots that meet the merging conditions in the spot change information set to obtain a spot information set to be merged.

[0092] Specifically, the spot screening unit 3 includes a spot centroid acquisition module 31 , a spot Voronoi diagram conversion module 32 and a spot Voronoi diagram screening module 33 .

[0093] See also Figure 3 , Figure 3 This is a simple schematic diagram of the corresponding centroid of the patch area. The slightly larger point in the figure is the centroid.

[0094] The patch centroid acquisition module 31 is used to execute step S31: process all patches in the patch change information set to obtain a patch information set.

[0095] Specifically, the centroid corresponding to the area of ​​the changed region in the spot change information set is calculated, and the centroid is saved in the spot change information set to obtain the spot information set. The spot is the centroid point corresponding to the area A of the changed region.

[0096] The centroid is the average point of all vertex positions of the polygon (or the mass concentration point of the polygon), that is, the center position of the polygon.

[0097] See also Figure 4 , Figure 4 This is a specific schematic diagram of the spatial segmentation calculation described in the present invention. The left figure is the centroid corresponding to the area of ​​the patch; the solid line in the middle figure is the triangular patch combination composed of the Delaunay triangulation, and the dotted line is the Voronoi diagram in the drawing process; the right figure is the completed Voronoi diagram, and each edge represents the adjacency relationship of the patches.

[0098] The spot Voronoi diagram conversion module 32 is used to execute step S32: perform spatial segmentation calculation on the spot information set to obtain the Voronoi diagram of the spot information set.

[0099] Specifically, a Delaunay triangulation network is formed by using the spots in the spot information set, and the spots and triangles are numbered to divide the spot combinations that constitute the triangles.

[0100] At the same time, the centers of the circumscribed circles of all triangles are calculated according to the triangle spot combination to obtain the circumcenter of the triangle. Then, the circumcenters of adjacent triangles with common edges are connected and saved in the Voronoi diagram linked list; the perpendicular bisector rays corresponding to the sides of triangles without common edges are saved in the Voronoi diagram linked list. After completing the traversal of the triangle spot combination, the Voronoi diagram of the spot information set is obtained.

[0101] The sides of the triangles without common edges can be represented as the edges of the Delaunay triangulation boundary, so it is only necessary to record the perpendicular bisector ray corresponding to the side into the Voronoi diagram linked list. The Voronoi diagram linked list is the data structure of the Voronoi diagram being drawn, that is, the Voronoi diagram that has not been completed.

[0102] Compared with the time complexity of querying adjacent objects in the patch information set, which is O(n 2 ), the time complexity of the present invention for generating a Voronoi diagram through a scanning line algorithm is O(nlogn), and the Voronoi diagram can better and quickly determine the connected objects in the space.

[0103] See also Figure 5 , Figure 5 It is a simple schematic diagram of the present invention for screening spots through the adjacency relationship of the Voronoi diagram, wherein the spot represented by each centroid belongs to class P (surface) in the diagram, and in the data structure, class P is composed of class C (line), and N is composed of class C (point).

[0104] The spot Voronoi diagram screening module 33 is used to execute step S33: traverse and screen the spot information set according to the adjacency relationship of the Voronoi diagram to obtain the spot set to be merged.

[0105] Specifically, according to the adjacent object relationship in the Voronoi graph, adjacent spots that are collinear and non-intersecting in the spot information set are screened out, and the adjacent spots that meet the requirements are retained, and the adjacent spots that do not meet the requirements are deleted from the spot information set to obtain the spot set to be merged.

[0106] Among them, the topological relationship can be used to quickly determine whether the blob corresponding to the current blob and all adjacent blob are completely adjacent but not overlapping on the boundary, so as to serve as candidate objects for merging; if the conditions are not met, it means that overlap may occur, and they are not candidates for merging.

[0107] The patch filtering unit 4 is used to execute step S4: performing slope filtering on the patch set to be merged according to elevation data to obtain a slope filtered patch set, and performing consistency check on the slope filtered patch set to obtain a patch set of elements of the same nature.

[0108] Specifically, the spot filtering unit 4 includes a spot slope filtering module 41 and a spot consistency filtering module 42 .

[0109] The patch slope filtering module 41 is used to execute step S41: performing slope analysis filtering on the patch set to be merged through elevation data to obtain a slope filtered patch set.

[0110] Specifically, quadratic surface fitting is performed on the landforms between the patches in the patch set to be merged to obtain the landform continuity, and interpolation analysis is performed based on the local landform continuity to obtain the local landform elevation difference. The patches with local landform elevation difference lower than a reference factor are retained, and the patches with elevation difference higher than a reference factor are deleted from the patch set to be merged to obtain the slope filtered patch set.

[0111] The quadratic surface fitting is to fit the terrain of the patch set to be merged by slope and aspect to simulate the continuous change of landforms and obtain landform continuity. The specific calculation method of the quadratic surface fitting is as follows:

[0112] The quadratic surface is calculated by a 3×3 moving window, the center of which is an elevation point e, as shown in Figure 7 As shown, based on this, the slope and aspect of the elevation point e are calculated as follows:

[0113]

[0114] Among them, dx is the slope change rate of elevation point e in the X direction, and dy is the slope change rate of elevation point e in the Y direction. The calculation method of dx and dy is as follows:

[0115]

[0116] Among them, e i is the elevation data i∈(1,9) around the elevation point e, such as Figure 7 As shown; X size and Y size The interval length of each elevation data in the 3×3 moving window is custom data.

[0117] Based on this, each elevation point is fitted by slope and aspect to obtain the landform continuity.

[0118] The interpolation analysis is to select key terrain points in multiple local areas of landform continuity as interpolation points according to elevation differences, so as to construct interpolation curves related to multiple areas, that is, landform elevation differences.

[0119] Based on this, the reference factor is set according to the maximum terrain slope in the local area, that is, the corresponding slope k is calculated according to the relevant interpolation curve in the local area, and it is directly judged whether the slope k exceeds the maximum slope grade standard of the current landform high-difference area, that is:

[0120] k≤tanα

[0121] Among them, α is the maximum slope angle of the current interpolation curve area.

[0122] Since the logical design of the reference factor is different according to different conditions, and this embodiment only provides a simple reference factor setting, the reference factor can also be logically extended according to different conditions. Therefore, the present invention does not specifically limit the specific setting content of the reference factor.

[0123] The slope is a measure of the ratio of height change in a specific area of ​​the ground; the aspect is the direction of the terrain slope, which is used to identify the downhill direction with the largest rate of change in the value from each pixel (elevation point) to its adjacent pixel (elevation point). Based on this, interpolation analysis can be used to further analyze and accurately identify terrain features and elevation changes based on fitting the quadratic surface through slope and aspect;

[0124] The interpolation analysis is mainly aimed at fitting a quadratic surface with slope and aspect, selecting characteristic points as interpolation points according to elevation differences, and constructing corresponding mathematical formulas to fit curves in the surface through mathematical formulas, and analyzing the curves. Since there are many specific algorithms for interpolation analysis, which are also common knowledge in the art, the present invention does not specifically limit the means of interpolation analysis.

[0125] See also Figure 6 , Figure 6 This is a simple schematic diagram of the third national land survey data.

[0126] The patch consistency filtering module 42 is used to execute step S42: performing consistency check on the slope filtering patch set to obtain a patch set of elements of the same nature.

[0127] Specifically, based on land survey data, it is determined whether the plots between the to-be-merged plots in the slope filtering plot set belong to the same type of plots in nature, and the plots that belong to the same type are stored in a plot set of elements with the same nature.

[0128] Among them, according to Figure 6 As shown, the land survey data is a geological planning classification disclosed by the state. Based on this, the present invention further analyzes the map patch merging according to the business data divided by the state to ensure that the merged map patches are more in line with business needs and to prevent the situation where agricultural land and commercial map patches are merged to be consistent.

[0129] The patch merging unit 5 is used to execute step S5: merging patches in the slope filtering patch set that conform to the same nature element patch set to obtain map data of complete segmented patches.

[0130] Among them, in the map data of the complete segmentation patch, the continuity of the landform is fully considered, that is, there will be no plot units where cliffs and plains intersect in a single patch at the same time. At the same time, the business attributes of the plot units are further considered, that is, adjacent plot units in a single patch will not have fields and forests at the same time, which ultimately makes the map data of the complete segmentation patch more logical.

[0131] Compared with the prior art, the present invention not only relies on the geometric properties of the spots themselves for merging, but also introduces the combined analysis and merging of geographic information data (DEM data) and business data (land survey data), which is more in line with actual geographical and business needs. In addition, the present invention further screens the merging conditions based on the slope and aspect in the geographic information data to ensure that the confidence and practicality of the merged spots are higher. At the same time, the present invention generates a Voronoi diagram through a scanning line algorithm to help quickly and accurately process large-scale map data (a large number of spots obtained by remote sensing image segmentation), and ultimately greatly optimizes the merging logic of the spots, and enhances the accuracy and efficiency of the analysis process.

[0132] Based on the same inventive concept, the present application also provides an electronic device, which may be a terminal device such as a server, a desktop computing device or a mobile computing device (e.g., a laptop computing device, a handheld computing device, a tablet computer, a netbook, etc.). The device includes one or more processors and a memory, wherein the processor is used to execute a program to implement the method for generating a pattern map based on spatial information according to an embodiment of the present invention; and the memory is used to store a computer program executable by the processor.

[0133] Based on the same inventive concept, the present application also provides a computer-readable storage medium, corresponding to the aforementioned embodiment of a method for generating a patch map based on spatial information, wherein the computer-readable storage medium stores a computer program thereon, and when the program is executed by a processor, the steps of the method for generating a patch map based on spatial information recorded in any of the aforementioned embodiments are implemented.

[0134] The present application may take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Computer-usable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be achieved by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0135] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, and the present invention is also intended to include these modifications and modifications.

Claims

1. A method for generating a spot map based on spatial information, characterized in that: The following steps are involved: S1: Acquire super-resolution remote sensing image data at different phases; S2: Performing spot change detection and spot semantic segmentation recognition on the preprocessed super-resolution remote sensing image data, and combining the detection results with the recognition results to obtain a spot change information set; S3: Traverse and filter out the adjacent spots that meet the merging conditions in the spot change information set to obtain the spot information set to be merged; wherein step S3 specifically includes the following sub-steps: S31: Process all spots in the spot change information set to obtain a spot information set; S32: performing spatial segmentation calculation on the image spot information set to obtain a Voronoi diagram of the image spot information set; The spatial segmentation calculation is specifically expressed as follows: The Delaunay triangulation network is formed by the spots in the spot information set, and the spots and triangles are numbered to divide the spot combinations that form the triangles; Calculate the circumcenters of all triangles according to the triangle spot combination to obtain the circumcenter of the triangle, then connect the circumcenters of adjacent triangles with common edges and save them in a one-dimensional Novo graph linked list; The perpendicular bisector rays corresponding to the sides of the triangles that do not have common sides are saved in the Voronoi diagram linked list. After the traversal is completed, the Voronoi diagram of the graph spot information set is obtained; S33: traverse and filter the spot information set according to the adjacency relationship of the Voronoi graph to obtain the spot set to be merged; S4: Slope filtering the to-be-merged patch set according to elevation data to obtain a slope-filtered patch set, and consistency checking the slope-filtered patch set to obtain a patch set of elements of the same nature; wherein the consistency check is to determine whether the plots between the to-be-merged patches in the slope-filtered patch set belong to the same type of plots in nature according to land survey data, and store the patches that belong to the same type in the patch set of elements of the same nature; S5: Merge the slope filtering patch set according to the patches of the feature patch set of the same nature to obtain map data of the complete segmented patches.

2. The method for generating a spot map based on spatial information according to claim 1, characterized in that: The step S2 comprises the following steps: S21: Slice and match the super-resolution remote sensing image data to obtain a number of tile pairs at different times and locations; S22A: performing convolution and contrast detection on a plurality of tile pairs in sequence, and calculating the area of ​​the detection results to obtain the change areas of the plurality of tile pairs; S22B: performing convolution feature extraction and semantic segmentation on a number of tile pairs to obtain semantic information of the number of tile pairs; S23: combining the changed regions of all tile pairs with the corresponding semantic information to obtain a patch change information set; Among them, the patch change information set includes the longitude and latitude boundary point set of the changed area, the area of ​​the changed area and the semantic change status of the changed area; the longitude and latitude boundary point set of the changed area is the longitude and latitude boundary point set P of the patch, the area of ​​the changed area is the area A of the changed patch corresponding to the longitude and latitude boundary point set P, and the semantic status of the changed area is the semantic information C of the tile pair corresponding to the area A of the changed patch.

3. The method for generating a spot map based on spatial information according to claim 2, characterized in that: The step S4 comprises the following steps: S41: performing slope analysis filtering on the patch set to be merged through elevation data to obtain a slope filtering patch set; The analysis and filtering is to perform quadratic surface fitting on the landforms between the spots in the set of spots to be merged to obtain landform continuity, and perform interpolation analysis based on the local landform continuity to obtain the local landform elevation difference, retain the spots whose local landform elevation difference is lower than a reference factor, and delete the spots whose local landform elevation difference exceeds a reference factor from the set of spots to be merged to obtain a slope filtered spot set; S42: Perform consistency check on the slope filtering patch set to obtain a patch set of elements with the same properties.

4. A device for generating a spot map based on spatial information, characterized in that: It includes a remote sensing image data acquisition unit, a multi-temporal spot change information acquisition unit, a spot screening unit, a spot filtering unit and a spot merging unit; The remote sensing image data acquisition unit is used to acquire super-resolution remote sensing image data of different phases; The multi-temporal spot change information acquisition unit is used to perform spot change detection and spot semantic segmentation recognition on the pre-processed super-resolution remote sensing image data, and combine the detection result and the recognition result to obtain a spot change information set; The spot screening unit is used to traverse and screen out adjacent spots that meet the merging conditions in the spot change information set to obtain the spot information set to be merged; wherein the spot screening unit also includes a spot centroid acquisition module, a spot Voronoi diagram conversion module and a spot Voronoi diagram screening module; The spot centroid acquisition module is used to process all spots in the spot change information set to obtain the spot information set; The image spot Voronoi diagram conversion module is used to perform spatial segmentation calculation on the image spot information set to obtain the Voronoi diagram of the image spot information set; The spatial segmentation calculation is specifically expressed as follows: The Delaunay triangulation network is formed by the spots in the spot information set, and the spots and triangles are numbered to divide the spot combinations that form the triangles; Calculate the circumcenters of all triangles according to the triangle spot combination to obtain the circumcenter of the triangle, then connect the circumcenters of adjacent triangles with common edges and save them in the Voronoi graph linked list; The perpendicular bisector rays corresponding to the sides of the triangles that do not have common sides are saved in the Voronoi diagram linked list. After the traversal is completed, the Voronoi diagram of the graph spot information set is obtained; The image spot Voronoi diagram screening module is used to traverse and screen the image spot information set according to the adjacency relationship of the Voronoi diagram to obtain the image spot set to be merged; The spot filtering unit is used to perform slope filtering on the spot set to be merged according to elevation data to obtain a slope filtered spot set, and to perform consistency check on the slope filtered spot set to obtain a same-nature element spot set; wherein the consistency check is to determine whether the plots between the spots to be merged in the slope filtered spot set belong to the same type of plots in nature according to land survey data, and store the spots that belong to the same type of plots in the same-nature element spot set; The patch merging unit is used to merge the patches in the slope filtering patch set that conform to the element patch set of the same nature to obtain map data of the complete segmented patches.

5. The device for generating a pattern map based on spatial information according to claim 4, characterized in that: The multi-temporal spot change information acquisition unit includes a remote sensing image slice matching module, a change area detection module, a semantic segmentation neural network and a spot change information combination module; The remote sensing image slice matching module is used to slice and match the super-resolution remote sensing image data to obtain a number of tile pairs that are co-located at different times; The change region detection module is used to perform convolution and contrast detection on a plurality of tile pairs in sequence, and calculate the region of the detection result to obtain the change region of the plurality of tile pairs; The semantic segmentation neural network is used to perform convolution feature extraction and semantic segmentation on a plurality of tile pairs to obtain semantic information of the plurality of tile pairs; The patch change information combining module is used to combine the change areas of all tile pairs with the corresponding semantic information to obtain a patch change information set; Among them, the patch change information set includes the longitude and latitude boundary point set of the changed area, the area of ​​the changed area and the semantic change status of the changed area; the longitude and latitude boundary point set of the changed area is the longitude and latitude boundary point set P of the patch, the area of ​​the changed area is the area A of the changed patch corresponding to the longitude and latitude boundary point set P, and the semantic status of the changed area is the semantic information C of the tile pair corresponding to the area A of the changed patch.

6. The device for generating a pattern map based on spatial information according to claim 5, characterized in that: The spot filtering unit includes a spot slope filtering module and a spot consistency filtering module; The patch slope filtering module is used to perform slope analysis filtering on the patch set to be merged through elevation data to obtain a slope filtered patch set; The analysis and filtering is to perform quadratic surface fitting on the landforms between the spots in the set of spots to be merged to obtain landform continuity, and perform interpolation analysis based on the local landform continuity to obtain the local landform elevation difference, retain the spots whose local landform elevation difference is lower than a reference factor, and delete the spots whose local landform elevation difference exceeds a reference factor from the set of spots to be merged to obtain a slope filtered spot set; The patch consistency filtering module is used to perform consistency check on the slope filtering patch set to obtain a patch set of elements of the same nature.

7. An electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for generating a patch map based on spatial information as described in any one of claims 1 to 3 is implemented.

8. A computer-readable storage medium storing computer-executable instructions, characterized in that: When the computer executable instructions are executed by the processor, a method for generating a patch map based on spatial information as described in any one of claims 1 to 3 is implemented.

Images