Method, device and equipment for determining change region of figure spot and storage medium

Abstract

The application provides a method and device for determining a change region of a map spot, equipment and a storage medium, wherein the method comprises: obtaining a to-be-processed remote sensing image of a preset region and a historical map spot of a target ground object type in a historical remote sensing image of the preset region, performing ground object classification on the to-be-processed remote sensing image to obtain a current map spot of the target ground object type, performing grid processing on the current map spot to obtain a first grid corresponding to the current map spot, performing grid processing on the historical map spot to obtain a second grid corresponding to the historical map spot, and determining a change region of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid. The change region of the map spot is determined based on a grid space, and the accuracy and determination efficiency of the change region of the map spot are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computers, in particular to a method and device for determining a change region of a map spot, an equipment and a storage medium. BACKGROUND

[0002] With the rapid development of remote sensing technology, remote sensing technology has been widely applied. For example, in agricultural production, remote sensing technology is often used to monitor cultivated land in all directions to identify changed cultivated land.

[0003] In the prior art, the map spot of a target land object type on a remote sensing image is usually identified, and then superimposed and analyzed with the historical map spot of the target land object type on a historical remote sensing image to determine the change region of the map spot of the target land object type.

[0004] However, the map spot extracted based on the image and the historical map spot often have the problem of misaligned map spot boundaries, and a large number of fragmented map spots will be obtained after superimposed analysis, resulting in low accuracy of the determined change region of the map spot and low efficiency of determining the change region of the map spot. SUMMARY

[0005] Therefore, the embodiments of the present application provide a method and device for determining a change region of a map spot, an equipment and a storage medium to solve the problem of low accuracy of the determined change region of the map spot and low efficiency of determining the change region of the map spot in the prior art.

[0006] In a first aspect, the embodiments of the present application provide a method for determining a change region of a map spot, comprising:

[0007] obtaining a to-be-processed remote sensing image of a preset region and a historical map spot of a target land object type in a historical remote sensing image of the preset region;

[0008] performing land object classification on the to-be-processed remote sensing image to obtain a current map spot of the target land object type;

[0009] performing grid processing on the current map spot to obtain a first grid corresponding to the current map spot, and performing grid processing on the historical map spot to obtain a second grid corresponding to the historical map spot;

[0010] determining a change region of a map spot of the target land object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid.

[0011] In an optional implementation, the grid processing on the current map spot to obtain the first grid corresponding to the current map spot comprises:

[0012] determining a target grid in which the current map spot is located from a preset vector grid space.

[0013] determining the first grid from the target grid according to the overlapping area of the target grid and the current polygon.

[0014] In an optional implementation, before the determining the polygon change area of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid, the method further comprises:

[0015] calculating the connected area of the connected grid in the first grid;

[0016] filtering the first grid according to the connected area of the connected grid in the first grid to obtain a filtered first grid;

[0017] The determining the polygon change area of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid comprises:

[0018] determining the polygon change area according to the filtered first grid and the second grid.

[0019] In an optional implementation, the determining the polygon change area according to the filtered first grid and the second grid comprises:

[0020] eroding the second grid to obtain an eroded second grid;

[0021] determining an overlapping grid between the filtered first grid and the eroded second grid;

[0022] determining the polygon change area according to the eroded second grid and the overlapping grid.

[0023] In an optional implementation, the determining the polygon change area according to the eroded second grid and the overlapping grid comprises:

[0024] determining an eroded change grid according to the eroded second grid and the overlapping grid;

[0025] performing inflation processing on the eroded change grid to obtain an inflated change grid;

[0026] determining a target change grid according to the inflated change grid and the second grid;

[0027] determining the polygon change area according to the target change grid.

[0028] In an optional embodiment, before the expanding the changed grid after the corrosion to obtain the expanded changed grid, the method further comprises:

[0029] calculating a connected area of a connected grid in the changed grid after the corrosion;

[0030] filtering the changed grid after the corrosion according to the connected area of the connected grid in the changed grid after the corrosion to obtain a filtered changed grid;

[0031] the expanding the changed grid after the corrosion to obtain the expanded changed grid comprises:

[0032] expanding the filtered changed grid to obtain the expanded changed grid.

[0033] In an optional embodiment, the classifying the to-be-processed remote sensing image to obtain the current map patch of the target ground object type comprises:

[0034] extracting feature maps of the to-be-processed remote sensing image by using a backbone network and a semantic context module to obtain a plurality of feature maps of the to-be-processed remote sensing image, the backbone network comprising a plurality of convolution layers and a semantic context module, an output of a previous convolution layer in the plurality of convolution layers being an input of a next convolution layer, and an output of a last convolution layer in the plurality of convolution layers being an input of the semantic context module;

[0035] performing feature fusion according to the plurality of feature maps by using a multi-scale fusion module to obtain a plurality of fused feature maps;

[0036] performing feature splicing according to the plurality of fused feature maps by using a decoding network to obtain a spliced feature map, and performing decoding processing according to the spliced feature map to obtain the current map patch.

[0037] In a second aspect, the embodiments of the present application further provide a determination device for a map patch change area, comprising:

[0038] an acquisition module configured to acquire a to-be-processed remote sensing image of a preset area and a historical map patch of a target ground object type in a historical remote sensing image of the preset area;

[0039] a processing module configured to classify the to-be-processed remote sensing image to obtain a current map patch of the target ground object type;

[0040] the processing module is further configured to perform grid processing on the current map patch to obtain a first grid corresponding to the current map patch, and perform grid processing on the historical map patch to obtain a second grid corresponding to the historical map patch;

[0041] determining, according to the first grid and the second grid, a region of change of the map patch of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image.

[0042] In an optional implementation, the processing module is specifically configured to:

[0043] determine a target grid in which the current map patch is located from a preset vector grid space;

[0044] determine the first grid from the target grid according to a coincidence area of the target grid and the current map patch.

[0045] In an optional implementation, the processing module is further configured to:

[0046] calculate a connected area of connected grids in the first grid;

[0047] filter the first grid according to the connected area of the connected grids in the first grid to obtain a filtered first grid;

[0048] The determining module is specifically configured to:

[0049] determine the region of change of the map patch according to the filtered first grid and the second grid.

[0050] In an optional implementation, the determining module is specifically configured to:

[0051] perform erosion processing on the second grid to obtain an eroded second grid;

[0052] determine a coincident grid between the filtered first grid and the eroded second grid;

[0053] determine the region of change of the map patch according to the eroded second grid and the coincident grid.

[0054] In an optional implementation, the determining module is specifically configured to:

[0055] determine an eroded change grid according to the eroded second grid and the coincident grid;

[0056] perform inflation processing on the eroded change grid to obtain an inflated change grid;

[0057] determine a target change grid according to the inflated change grid and the second grid;

[0058] determine the region of change of the map patch according to the target change grid.

[0059] In an optional implementation, the processing module is further configured to:

[0060] calculate a connected area of connected grids in the changed grid after the corrosion;

[0061] filter the changed grid after the corrosion according to the connected area of the connected grids in the changed grid after the corrosion, to obtain a filtered changed grid;

[0062] The processing module is specifically configured to:

[0063] perform inflation processing on the filtered changed grid, to obtain the inflated changed grid.

[0064] In an optional implementation, the processing module is specifically configured to:

[0065] extract feature maps from the to-be-processed remote sensing image by using a backbone network and a semantic context module, to obtain a plurality of feature maps of the to-be-processed remote sensing image, the backbone network comprises a plurality of convolution layers and one semantic context module, an output of a previous convolution layer in the plurality of convolution layers is an input of a next convolution layer, and an output of a last convolution layer in the plurality of convolution layers is an input of the semantic context module;

[0066] perform feature fusion according to the plurality of feature maps by using a multi-scale fusion module, to obtain a plurality of fused feature maps;

[0067] perform feature splicing according to the plurality of fused feature maps by using a decoding network, to obtain a spliced feature map, and perform decoding processing according to the spliced feature map, to obtain the current graph patch.

[0068] In a third aspect, an electronic device is provided, which comprises a processor, a memory and a bus, the memory stores machine readable instructions executable by the processor, when the electronic device is running, the processor and the memory communicate through the bus, and the processor executes the machine readable instructions to perform the method for determining a graph patch change area according to any one of the first aspect.

[0069] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, when the computer program is run by a processor, the method for determining a graph patch change area according to any one of the first aspect is executed.

[0070] The application provides a determination method, device and equipment of a graph spot change area and a storage medium, and the method comprises the following steps: acquiring a to-be-processed remote sensing image of a preset area and a historical graph spot of a target ground object type in a historical remote sensing image of the preset area, performing ground object classification on the to-be-processed remote sensing image to obtain a current graph spot of the target ground object type, performing grid processing on the current graph spot to obtain a first grid corresponding to the current graph spot, performing grid processing on the historical graph spot to obtain a second grid corresponding to the historical graph spot, and determining a graph spot change area of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid. The graph spot change area is determined based on a grid space, and the accuracy and determination efficiency of the graph spot change area are improved.

[0071] In order to make the above objectives, features and advantages of the present application more apparent, the following will describe a preferred embodiment in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation to the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0073] Figure 1 Flowchart of the determination method of the graph spot change area provided by the embodiments of the present application Figure 1 ;

[0074] Figure 2 Flowchart of the determination method of the graph spot change area provided by the embodiments of the present application Figure 2 ;

[0075] Figure 3 Schematic diagram of the preset vector grid space provided by the embodiments of the present application

[0076] Figure 4 Flowchart of the determination method of the graph spot change area provided by the embodiments of the present application Figure 3 ;

[0077] Figure 4 Flowchart of the determination method of the graph spot change area provided by the embodiments of the present application Figure 5 ;

[0078] Figure 5 Flowchart of the determination method of the graph spot change area provided by the embodiments of the present application Figure 6 ;

[0079] Figure 6A schematic diagram of a feature classification model of a target feature type provided for an embodiment of the present application;

[0080] Figure 7 A schematic diagram of an implementation process of a specific target change grid provided for an embodiment of the present application;

[0081] Figure 8 A structural schematic diagram of a determination device of a graph change area provided for an embodiment of the present application;

[0082] Figure 9 A structural schematic diagram of an electronic device provided for an embodiment of the present application. DETAILED DESCRIPTION

[0083] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the following will combine the accompanying drawings for the embodiments of the present application to make a clear and complete description of the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0084] After the overlay analysis, a large number of fine graph vector (may be caused by edge offset, mistaken for a graph change area) is obtained, resulting in a large difference between the result and the actual change (i.e. the accuracy of the determined graph change area is low), and in terms of calculation timeliness, the vector overlay analysis algorithm has a problem of low calculation efficiency in processing a large number of graph calculations, resulting in a long time to obtain the final result. Based on this, the present application provides a determination method of a graph change area, which extracts the graph of the concerned feature type by classifying the features of the remote sensing image, then maps the historical graph and the extracted graph to a grid space, processes the area boundary deviation in the grid space to determine the change grid corresponding to the two graphs, and then determines the graph change area based on the change grid, thereby improving the accuracy and determination efficiency of the graph change area.

[0085] The following will be described in combination with Figure 10 - Figure 1 The determination method of the graph change area provided by the present application is described.

[0086] Figure 8 A flowchart of the determination method of the graph change area provided for an embodiment of the present application Figure 1The execution subject of the embodiment can be an electronic device, such as a computer, a mobile phone, a server, or the like.

[0087] As shown in the method can include: Figure 1

[0088] S101, obtaining a to-be-processed remote sensing image of a preset area, and a historical map patch of a target ground object type in a historical remote sensing image of the preset area.

[0089] Generally, taking a topographic map as a working base map, land units with the same landform, land use type, and soil erosion type are classified into one category, which is taken as a basic investigation unit, and then the unit is drawn on the topographic map to become a map patch.

[0090] The preset area is monitored by using a remote sensing technology to obtain a to-be-processed remote sensing image of the preset area and a historical remote sensing image, wherein the acquisition time of the historical remote sensing image is earlier than that of the to-be-processed remote sensing image, and the historical map patch of the target ground object type in the historical remote sensing image can be understood as a map patch of the target ground object type in the historical remote sensing image, and the target ground object type can be, for example, cultivated land or a building.

[0091] S102, performing ground object classification on the to-be-processed remote sensing image to obtain a current map patch of the target ground object type.

[0092] Ground object classification refers to classifying objects on the ground in the to-be-processed remote sensing image, and the current map patch of the target ground object type is determined from the to-be-processed remote sensing image by performing ground object classification on the to-be-processed remote sensing image, wherein the current map patch can be understood as a map patch of the target ground object type in the to-be-processed remote sensing image.

[0093] It is worth noting that the historical map patch of the target ground object type in the historical remote sensing image can be obtained by performing ground object classification on the historical remote sensing image in advance, and the map patch that meets the actual situation is confirmed by manual confirmation.

[0094] S103, performing grid processing on the current map patch to obtain a first grid corresponding to the current map patch, and performing grid processing on the historical map patch to obtain a second grid corresponding to the historical map patch.

[0095] The current map patch is mapped to a grid space, the current map patch is grid processed to obtain a first grid corresponding to the current map patch, and similarly, the historical map patch is mapped to a grid space, the historical map patch is grid processed to obtain a second grid corresponding to the historical map patch.

[0096] ​In some embodiments, a meshing processing algorithm can be employed to map the current patch according to patch information of the current patch, convert the current patch to a mesh space, and obtain a first mesh corresponding to the current patch, wherein the patch information of the current patch includes but is not limited to the geometric shape of the current patch and geographic coordinates, which can be the latitude and longitude coordinates of each point on the current patch in the real scene.

[0097] Similarly, a meshing processing algorithm can be employed to map the historical patch according to patch information of the historical patch, convert the historical patch to a mesh space, and obtain a second mesh corresponding to the historical patch, wherein the patch information of the historical patch includes but is not limited to the geometric shape of the historical patch and geographic coordinates, which can be the latitude and longitude coordinates of each point on the historical patch in the real scene.

[0098] S104, determining a patch change region of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first mesh and the second mesh.

[0099] In actual application, in order to monitor the patch change of the target ground object type in the preset region, for example, to monitor the patch change of the farmland in the preset region, so as to determine the decrease or increase of the farmland in the preset region, the current patch and the historical patch of the target ground object type in the preset region can be obtained, and the current patch and the historical patch are mapped to the mesh space respectively to obtain the first mesh corresponding to the current patch and the second mesh corresponding to the historical patch, and then the change mesh between the first mesh and the second mesh is determined according to the first mesh and the second mesh, which can be the decrease mesh and the increase mesh, wherein the decrease mesh can be understood as the mesh decreased by the second mesh relative to the first mesh, and the increase mesh can be understood as the mesh increased by the second mesh relative to the first mesh.

[0100] Then the change mesh can be converted to obtain a change region corresponding to the change mesh, wherein the change region is the patch change region of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image, wherein the patch change region can be understood as the region of the patch decreased or increased by the current patch relative to the historical patch.

[0101] It is worth noting that after the patch change region is determined, the to-be-processed remote sensing image can also be rendered according to the patch change region to obtain a rendered to-be-processed remote sensing image, so that the change patch of the target ground object type can be directly viewed from the rendered to-be-processed remote sensing image.

[0102] In the method for determining the region of change of the map spot in the embodiment, the current map spot of the target ground object type is obtained by performing ground object classification on the remote sensing image to be processed, the first grid corresponding to the current map spot is obtained by performing grid processing on the current map spot, and the second grid corresponding to the historical map spot is obtained by performing grid processing on the historical map spot. The region of change of the map spot of the target ground object type between the remote sensing image to be processed and the historical remote sensing image is determined according to the first grid and the second grid. The region of change of the map spot is determined based on the grid space, and the accuracy and determination efficiency of the region of change of the map spot are improved.

[0103] Figure 2 Flowchart of the method for determining the region of change of the map spot provided in the embodiment Figure 2 As shown in Figure 2 In an optional embodiment, step S103, the current map spot is grid processed to obtain the first grid corresponding to the current map spot, can include:

[0104] S201, determining the target grid in which the current map spot is located from the preset vector grid space.

[0105] The preset vector space is divided into grids to obtain the preset vector grid space, and the current map spot is mapped to the preset vector grid space to determine the target grid in which the current map spot is located from the preset vector grid space, wherein the target grid can be the grid in which the current map spot is located in the preset vector grid space.

[0106] S202, determining the first grid from the target grid according to the overlapping area of the target grid and the current map spot.

[0107] The number of target grids is multiple, the overlapping area of the target grid and the current map spot is calculated, if the overlapping area meets the preset area condition, the corresponding target grid is determined as the first grid, if the overlapping area does not meet the preset area condition, the corresponding target grid is filtered out, wherein the preset area condition can be, for example, exceeding a preset area threshold or exceeding 50% of the area of a grid in the preset vector grid space, which is not particularly limited in the embodiment.

[0108] In some embodiments, the preset vector space is divided into 1m*1m grids to obtain the preset vector grid space, the entire grid can be represented as A[m][n] by a matrix of length m and width n, if 50% of the area of the grid at position (i,j) is in the target grid, the value of matrix element A[i][j] is set to 1, otherwise the value is set to 0.

[0109] Similarly, performing gridding on historical patches to obtain the second grid corresponding to the historical patches may include: determining the grid where the historical patch is located from a preset vector grid space, and determining the second grid from the grid where the historical patch is located based on the overlapping area between the grid where the historical patch is located and the current patch.

[0110] The process of determining the second grid is similar to that of the first grid. For details, please refer to the above description of the process of determining the first grid. It will not be repeated here.

[0111] Figure 3 A schematic diagram of the preset vector grid space provided in the embodiments of this application, as shown below. Figure 3 As shown, it includes triangular and trapezoidal features, and then the first grid determined from the target grid by the overlapping area is marked as the grid enclosed in dashed lines.

[0112] In the method for determining the change of a patch in this embodiment, the target grid where the current patch is located is determined from a preset vector grid space. Based on the overlapping area between the target grid and the current patch, a first grid is determined from the target grid. By processing the boundary deviation of the current patch in the grid space and filtering by area, deviations in the extracted first grid due to boundary offset are prevented, which helps to improve the accuracy of the patch change area.

[0113] Figure 4 A flowchart illustrating the method for determining the region of image variation provided in this application embodiment. Figure 3 ,like Figure 4 As shown, in an optional implementation, before step S104, which determines the area of ​​change in target land cover type between the remote sensing image to be processed and historical remote sensing images based on the first grid and the second grid, the method may further include:

[0114] S301. Calculate the connected area of ​​the connected grids in the first grid.

[0115] S302. Filter the first grid based on the connected area of ​​the connected grids in the first grid to obtain the filtered first grid.

[0116] by Figure 5 For example, grids 1, 2, and 3 are connected grids. The connected area of ​​the connected grids in the first grid is calculated, where the connected area is the area of ​​the connected grids in the first grid. Then, the first grid is filtered based on the connected area of ​​the connected grids in the first grid to obtain the filtered first grid. Connected grids with connected areas less than a preset area threshold can be filtered out. The preset area threshold can be, for example, 50. That is, for the first grid, connected regions with connected areas less than 50 are filtered out to obtain the filtered first grid.

[0117] It is worth noting that the preset area threshold can be determined according to the target ground object type. In order to filter the grid which is less likely to be the target ground object type, the first grid can be filtered by the connected area to obtain the filtered first grid. For example, for farmland, the area less than 50 is less likely to be farmland, and for buildings, the area less than 40 is less likely to be buildings.

[0118] Correspondingly, in step S103, the graph change area of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image is determined according to the first grid and the second grid, which can include:

[0119] S303, determining the graph change area according to the filtered first grid and the second grid.

[0120] The first grid is filtered by the connected area to obtain the filtered first grid, and then the filtered first grid and the second grid are determined according to the filtered first grid and the second grid, and the graph change area between the to-be-processed remote sensing image and the historical remote sensing image is determined according to the change grid.

[0121] In an optional embodiment, in step S303, the graph change area is determined according to the filtered first grid and the second grid, which can include:

[0122] The second grid is eroded to obtain the eroded second grid, the overlapping grid between the filtered first grid and the eroded second grid is determined, and the graph change area is determined according to the eroded second grid and the overlapping grid.

[0123] The second grid is eroded to obtain the eroded second grid, and then the overlapping grid between the filtered first grid and the eroded second grid is determined, wherein the corresponding positions of the matrix corresponding to the filtered first grid and the matrix corresponding to the eroded second grid can be multiplied to obtain the matrix corresponding to the overlapping grid. In this way, the overlapping grid can be determined according to the matrix corresponding to the overlapping grid, and the overlapping grid is the same grid between the filtered first grid and the eroded second grid.

[0124] Then, according to the second grid after erosion and the coincident grid, a grid other than the coincident grid is determined from the second grid after erosion, and a planar patch change region is determined according to the grid. In this way, a difference operation can be performed on a matrix corresponding to the second grid after erosion and a matrix corresponding to the coincident grid to obtain a matrix corresponding to the grid other than the coincident grid in the second grid after erosion. Then, the grid other than the coincident grid in the second grid after erosion can be determined according to the matrix corresponding to the grid other than the coincident grid in the second grid after erosion. Further, the planar patch change region can be determined according to the grid other than the coincident grid in the second grid after erosion.

[0125] It can be understood that, then, for the first grid after filtering, the grid reduced relative to the second grid after erosion is the grid other than the coincident grid in the second grid after erosion. In this way, by using the second grid corresponding to the historical planar patch after erosion, deviation of the second grid due to boundary offset can be prevented, and the accuracy of the planar patch change region can be improved.

[0126] The above-described erosion algorithm can be implemented by using a circular kernel with a kernel diameter of 30. For implementation of the erosion algorithm, refer to related descriptions in the prior art, which will not be repeated here.

[0127] In the method for determining the planar patch change region in this embodiment, the first grid is filtered according to the connected area of the connected grid in the first grid to obtain a first grid after filtering, and the planar patch change region is determined according to the first grid after filtering and the second grid. By filtering the first grid according to the connected area, the grid that does not meet the target ground object type can be excluded, and the accuracy of the planar patch change region can be improved.

[0128] Figure 5 Flowchart of the method for determining the planar patch change region provided in this embodiment Figure 6 As shown in FIG. 1, in an optional embodiment, the planar patch change region is determined according to the second grid after erosion and the coincident grid, and can include the following steps. Figure 5 S401, a second grid after erosion and a coincident grid are determined.

[0129] According to the second grid after erosion and the coincident grid, a grid other than the coincident grid is determined from the second grid after erosion, and the grid is determined as a change grid after erosion. In this way, a difference operation can be performed on a matrix corresponding to the second grid after erosion and a matrix corresponding to the coincident grid to obtain a matrix corresponding to the change grid after erosion. Then, the change grid after erosion can be determined according to the matrix corresponding to the change grid after erosion.

[0130]

[0131] ​S402, performing inflation processing on the changed grid after corrosion to obtain an inflated changed grid after corrosion.

[0132] After the changed grid after corrosion is determined, inflation processing can be performed on the changed grid after corrosion to obtain an inflated changed grid after corrosion. Since the changed grid after corrosion is obtained after corrosion of the second grid, the changed grid after corrosion can not include all the changed grids between the filtered first grid and the second grid. Therefore, inflation processing is performed on the changed grid after corrosion to make the inflated changed grid after corrosion include all the changed grids between the filtered first grid and the second grid as much as possible.

[0133] The inflation algorithm can be implemented by using a circular kernel with a kernel diameter of 30. For the implementation process of the inflation algorithm, refer to the related description of the prior art, which will not be repeated here.

[0134] In an optional embodiment, before step S402, the method can further include:

[0135] The connected area of the connected grid in the changed grid after corrosion is calculated, and the changed grid after corrosion is filtered according to the connected area of the connected grid in the changed grid after corrosion to obtain a filtered changed grid.

[0136] Correspondingly, step S402, performing inflation processing on the changed grid after corrosion to obtain an inflated changed grid after corrosion, includes: performing inflation processing on the filtered changed grid to obtain an inflated changed grid after corrosion.

[0137] Since the changed grid after corrosion with a too small connected area is less likely to be a changed grid of the target ground object type (i.e., low credibility), the connected area of the connected grid in the changed grid after corrosion can be calculated, and the changed grid after corrosion is filtered according to the connected area of the connected grid in the changed grid after corrosion to obtain a filtered changed grid. The connected area that does not meet a preset area condition is filtered, and the filtered changed grid is then inflated to obtain an inflated changed grid after corrosion. The preset area condition can be, for example, an area less than 50.

[0138] S403, determining a target changed grid according to the inflated changed grid after corrosion and the second grid.

[0139] Since the inflated changed grid after corrosion can include part of the grid in the second grid, the target changed grid can be determined according to the inflated changed grid after corrosion and the second grid, the overlapping grid between the second grid and the inflated changed grid after corrosion, and the grid in the inflated changed grid after corrosion except the overlapping grid, as follows:

[0140] D[i][j] = Dn[i][j] - Dn[i][j] * B[i][j], D[i][j] is a matrix corresponding to the expanded change grid, B[i][j] is a matrix corresponding to the second grid, and Dn[i][j] * B[i][j] is a matrix corresponding to the overlapping grid between the second grid and the expanded change grid.

[0141] S404, determining the planimetric change area according to the target change grid.

[0142] The grid vectorization algorithm is used to convert the target change grid into a planimetric change vector, so as to obtain the planimetric change area.

[0143] In the planimetric change area determination method of the embodiment, the change grid after erosion is determined according to the second grid after erosion and the overlapping grid, the target change grid is determined according to the expanded change grid and the second grid, and the planimetric change area is determined according to the target change grid. Through erosion and expansion, the deviation of the extracted grid due to boundary offset is prevented, which helps to improve the accuracy of the planimetric change area.

[0144] Figure 6 The flowchart of the planimetric change area determination method provided by the embodiment of the application is shown in Figure 7 As shown in Figure 7 In an optional embodiment, step S102, classifying the ground objects of the to-be-processed remote sensing image to obtain the current planimetric of the target ground object type, can include:

[0145] S501, using a backbone network and a semantic context module to extract feature maps of the to-be-processed remote sensing image to obtain a plurality of feature maps of the to-be-processed remote sensing image.

[0146] The backbone network includes a plurality of convolution layers and a semantic context module, the output of the previous convolution layer in the plurality of convolution layers is the input of the next convolution layer, and the output of the last convolution layer in the plurality of convolution layers is the input of the semantic context module.

[0147] Each convolution layer outputs one feature map, and the semantic context module outputs one feature map.

[0148] The semantic context module can be implemented using Atrous Spatial Pyramid Pooling (ASPP), which increases the receptive field to capture semantic context information. In deep convolutional networks, the size of the receptive field roughly reflects the amount of context information acquired. However, in practice, the receptive field of high-level models is very small, resulting in a limited range of features learned by the model. The ASPP module can increase the receptive field without generating unnecessary parameters and can also improve the model's segmentation accuracy.

[0149] Figure 7 A schematic diagram of the land cover classification model for the target land cover type provided in the embodiments of this application, as shown below. Figure 7 As shown, the model includes a decoder, a multi-scale fusion module, and a decoder. The decoder includes a backbone network and an ASPP. The multi-level convolutional layers of the backbone network are L1, L2, L3, and L4. The first-level convolutional layer (L1) is used to extract feature maps from the remote sensing image to be processed, obtaining the first feature map of the remote sensing image to be processed. Then, the first feature map is used as the input of the second-level convolutional layer (L2), which outputs the second feature map. This process is repeated to obtain the third feature map, the fourth feature map, and the fifth feature map output by the ASPP.

[0150] S502. A multi-scale fusion module is used to fuse features based on multiple feature maps to obtain multiple fused feature maps.

[0151] A multi-scale fusion module is used to fuse features from multiple feature maps, resulting in multiple fused feature maps. See [link to module]. Figures 1-7 A bidirectional pyramid grid can be used to aggregate feature maps at different levels, extract feature maps at levels L2-L4, and use bidirectional feature fusion from top to bottom and bottom to top to obtain multiple fused feature maps. For the first pyramid (i.e., the top-down pyramid), the fifth feature map and the third feature map are fused to obtain the first fused feature map, and the first fused feature map and the second feature map are fused to obtain the second fused feature map. For the second pyramid (i.e., the bottom-up pyramid), the first feature map and the second fused feature map are fused to obtain the third fused feature map, and the third fused feature map, the second fused feature map, and the second feature map are fused to obtain the fourth fused feature map. The fourth fused feature map, the first fused feature map, and the third feature map are fused to obtain the fifth fused feature map, and the fifth fused feature map and the fourth feature map are fused to obtain the sixth fused feature map.

[0152] S503. A decoding network is used to stitch together features from multiple fused feature maps to obtain a stitched feature map, and the stitched feature map is then decoded to obtain the current patch.

[0153] The decoding network is used to splice the plurality of fused feature maps first, and then decode the spliced feature map to obtain the current patch, see Figure 8 The decoding network is used to splice the fourth feature map, the fifth fused feature map, the fourth fused feature map and the third fused feature map through a decoder to obtain a spliced feature map, which includes the features of the fourth feature map, the fifth fused feature map, the fourth fused feature map and the third fused feature map, and then decode the spliced feature map to obtain the current patch of the target ground object type of the to-be-processed remote sensing image, wherein the spliced feature map carries the local and global context information of the to-be-processed remote sensing image.

[0154] It can be understood that, in feature fusion, the dimensions of the to-be-fused feature maps can be sampled to the same size (i.e., consistent dimensions) through convolution and upsampling, and then the sampled feature maps are summed to perform feature fusion. In feature splicing, the dimensions of the to-be-spliced feature maps can be sampled to the same size through convolution and upsampling, and then the sampled feature maps are spliced in a concat connection manner to obtain a spliced feature map, and then the spliced feature map is decoded to obtain the prediction value of each pixel point in the current patch, for example, the prediction value of a cultivated land pixel is 0 and the prediction value of a building pixel point is 1. By connecting the pixel points of the target ground object type, the current patch of the target ground object type can be determined.

[0155] It is worth noting that the above ground object classification model can be obtained by training. In the training process, a plurality of sample remote sensing images can be obtained, each sample remote sensing image is labeled with a corresponding patch, and the above ground object classification model is obtained by training the plurality of sample remote sensing images.

[0156] In the patch change area determination method of the embodiment, the ground object classification model of the remote sensing image is used to classify the input image (i.e., semantic segmentation), and the patch of the target ground object type is extracted. Since the model fuses multi-scale features, the extracted patch performs better in detail.

[0157] Based on the above, Figure 8 The patch change area determination method of the present application will be described below in conjunction with a specific embodiment.

[0158] Figure 9 A specific implementation process of the target change grid provided by the embodiment of the present application is shown in Figure 9 The first grid and the second grid are used as inputs to obtain the target change grid by using a change area analysis algorithm, and the specific algorithm process is as follows:

[0159] Step one, filter the connected region with connectivity less than 50 in the first grid Ac to obtain the filtered first grid Af.

[0160] Step two, use the corrosion algorithm (set the kernel as a circular kernel with a diameter of 30) to process the second grid B to obtain the corroded second grid Be.

[0161] Step three, multiply the corresponding positions of the Af matrix and the Be matrix to obtain the same region grid C.

[0162] Step four, use the Be matrix and the C matrix to calculate the difference at the same position to obtain the corroded change grid De.

[0163] Step five, filter the connected region with connectivity less than 20 in the corroded change grid De to obtain the filtered change corrosion grid Def.

[0164] Step six, use the dilatation algorithm (set the kernel as a circular kernel with a diameter of 30) to process the filtered change corrosion grid Def to obtain the restored change region grid Dn.

[0165] Step seven, use D[i][j] = Dn[i][j] - Dn[i][j] * B[i][j] to calculate the final target change grid D.

[0166] Step eight, use the grid vectorization algorithm to convert the target change grid into a change region vector to obtain the final graph change region.

[0167] In this scheme, a feature classification model for ground objects in remote sensing images is proposed, which integrates multi-scale features and extracts details of the graph better. In addition, a change region analysis algorithm based on grid space is used instead of a vector-based overlay analysis algorithm, which solves the problems of fine change graph and low calculation efficiency. For farming, the change of cultivated land based on satellite remote sensing images is used to replace the original manual visual interpretation to monitor the change of cultivated land.

[0168] Based on the same inventive concept, the embodiments of the present application also provide a graph change region determination device corresponding to the graph change region determination method. Since the principle of the device in the embodiments of the present application solves the problem, the implementation of the device can be referred to the implementation of the method, and the repeated parts will not be described here.

[0169] Figure 10 The structure diagram of the graph change region determination device provided by the embodiments of the present application is shown in the figure. Figure 10 As shown in the figure, the device can include:

[0170] The acquisition module 601 is configured to acquire a to-be-processed remote sensing image of a preset region and a historical map patch of a target ground object type in a historical remote sensing image of the preset region.

[0171] The processing module 602 is configured to perform ground object classification on the to-be-processed remote sensing image to obtain a current map patch of the target ground object type.

[0172] The processing module 602 is further configured to perform grid processing on the current map patch to obtain a first grid corresponding to the current map patch, and perform grid processing on the historical map patch to obtain a second grid corresponding to the historical map patch.

[0173] The determination module 603 is configured to determine a map patch change region of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid.

[0174] In an optional implementation, the processing module 602 is specifically configured to:

[0175] determine a target grid in which the current map patch is located from a preset vector grid space;

[0176] determine the first grid from the target grid according to an overlapping area of the target grid and the current map patch.

[0177] In an optional implementation, the processing module 602 is further configured to:

[0178] calculate a connected area of connected grids in the first grid;

[0179] perform filtering on the first grid according to the connected area of the connected grids in the first grid to obtain a filtered first grid;

[0180] The determination module 603 is specifically configured to:

[0181] determine the map patch change region according to the filtered first grid and the second grid.

[0182] In an optional implementation, the determination module 603 is specifically configured to:

[0183] perform erosion processing on the second grid to obtain an eroded second grid;

[0184] determine an overlapping grid between the filtered first grid and the eroded second grid;

[0185] determine the map patch change region according to the eroded second grid and the overlapping grid.

[0186] In an optional implementation, the determination module 603 is specifically configured to:

[0187] determine an eroded change grid according to the eroded second grid and the overlapping grid.

[0188] performing inflation processing on the changed grid after the corrosion to obtain an inflated changed grid;

[0189] determining a target changed grid according to the inflated changed grid and the second grid;

[0190] determining a region of the polygon change according to the target changed grid.

[0191] In an optional implementation, the processing module 602 is further configured to:

[0192] calculate a connected area of the connected grid in the changed grid after the corrosion;

[0193] perform filtering on the changed grid after the corrosion according to the connected area of the connected grid in the changed grid after the corrosion to obtain a filtered changed grid;

[0194] The processing module 602 is specifically configured to:

[0195] perform inflation processing on the filtered changed grid to obtain an inflated changed grid.

[0196] In an optional implementation, the processing module 602 is specifically configured to:

[0197] extract feature maps of the to-be-processed remote sensing image by using a backbone network and a semantic context module to obtain a plurality of feature maps of the to-be-processed remote sensing image, the backbone network includes a plurality of convolution layers and a semantic context module, an output of a previous convolution layer in the plurality of convolution layers is an input of a next convolution layer, and an output of a last convolution layer in the plurality of convolution layers is an input of the semantic context module;

[0198] perform feature fusion according to the plurality of feature maps by using a multi-scale fusion module to obtain a plurality of fused feature maps;

[0199] perform feature splicing according to the plurality of fused feature maps by using a decoding network to obtain a spliced feature map, and perform decoding processing according to the spliced feature map to obtain a current polygon.

[0200] The description of the processing procedure of each module in the device and the interaction procedure between the modules can refer to the related description in the foregoing method embodiments, and will not be described in detail here.

[0201] ​ A structural schematic diagram of an electronic device provided in the embodiments of the present application is shown in FIG. 1. ​As shown, the device can include a processor 701, a memory 702 and a bus 703, the memory 702 stores machine readable instructions executable by the processor 701, when the electronic device is running, the processor 701 communicates with the memory 702 through the bus 703, the processor 701 executes the machine readable instructions to perform the above-mentioned determination method of the graph change area.

[0202] The embodiment of the present application further provides a computer readable storage medium, the computer readable storage medium stores a computer program, the computer program is executed by a processor to perform the above-mentioned determination method of the graph change area.

[0203] In the embodiment of the present application, the computer program executed by the processor can also execute other machine readable instructions to perform the method as described in the embodiment, for the specific method steps and principles executed, refer to the description of the embodiment, and will not be described in detail here.

[0204] In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented by other means. The apparatus embodiment described above is only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some communication interface, and can be electrical, mechanical or other forms.

[0205] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment.

[0206] In addition, each functional unit in the embodiments provided in the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0207] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0208] It should be noted that similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In addition, the terms "first", "second", "third" and the like are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0209] Finally, it should be noted that the above-described embodiments are only specific implementations of the present application, which are used to illustrate the technical solutions of the present application, but not to limit them. The protection scope of the present application is not limited thereto. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily think of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed by the present application, or make equivalent replacements to some of the technical features. These modifications, changes or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application. They should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for determining a change area of a map spot, characterized by, The method comprises the following steps: acquiring a to-be-processed remote sensing image of a preset region and a historical image patch of a target ground object type in a historical remote sensing image of the preset region; performing ground object classification on the to-be-processed remote sensing image to obtain a current image patch of the target ground object type; performing grid processing on the current image patch to obtain a first grid corresponding to the current image patch, and performing grid processing on the historical image patch to obtain a second grid corresponding to the historical image patch; determining a patch change region of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid; before the step of determining the patch change region of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid, the method further comprises the following steps: calculating a connected area of connected grids in the first grid; filtering the first grid according to the connected area of the connected grids in the first grid to obtain a filtered first grid; the step of determining the patch change region of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid comprises the following step: determining the patch change region according to the filtered first grid and the second grid.

2. The method of claim 1, wherein, the step of performing grid processing on the current image patch to obtain the first grid corresponding to the current image patch comprises the following steps: determining a target grid in which the current image patch is located from a preset vector grid space; determining the first grid from the target grid according to an overlapping area of the target grid and the current image patch.

3. The method of claim 1, wherein, the step of determining the patch change region according to the filtered first grid and the second grid comprises the following steps: performing erosion processing on the second grid to obtain an eroded second grid; determining overlapping grids between the filtered first grid and the eroded second grid; determining the patch change region according to the eroded second grid and the overlapping grids.

4. The method of claim 3, wherein, the step of determining the patch change region according to the eroded second grid and the overlapping grids comprises the following steps: determining an eroded change grid according to the eroded second grid and the overlapping grids; performing inflation processing on the eroded change grid to obtain an inflated change grid; determining a target change grid according to the inflated change grid and the second grid; determining the patch change region according to the target change grid.

5. The method of claim 4, wherein, before the step of performing inflation processing on the eroded change grid to obtain the inflated change grid, the method further comprises the following steps: calculating a connected area of connected grids in the eroded change grid; filtering the eroded change grid according to the connected area of the connected grids in the eroded change grid to obtain a filtered change grid; the step of performing inflation processing on the eroded change grid to obtain the inflated change grid comprises the following step: performing inflation processing on the filtered change grid to obtain the inflated change grid.

6. The method of claim 1, wherein, The ground object classification on the to-be-processed remote sensing image obtains the current map patch of the target ground object type, and the method comprises the steps of: A backbone network and a semantic context module are used to extract feature maps of the to-be-processed remote sensing image, and a plurality of feature maps of the to-be-processed remote sensing image are obtained, the backbone network comprises a plurality of convolution layers and a semantic context module, the output of a previous convolution layer in the plurality of convolution layers is the input of a subsequent convolution layer, and the input of the semantic context module is the output of a last convolution layer in the plurality of convolution layers; A multi-scale fusion module is used to perform feature fusion according to the plurality of feature maps, and a plurality of fused feature maps are obtained; A decoding network is used to perform feature splicing according to the plurality of fused feature maps, and a spliced feature map is obtained, and decoding processing is performed according to the spliced feature map, and the current map patch is obtained.

7. A device for determining the region of change in a map feature, characterized in that, Comprise: The acquisition module is used for acquiring a to-be-processed remote sensing image of a preset area and a historical map patch of a target ground object type in a historical remote sensing image of the preset area; The processing module is used for performing ground object classification on the to-be-processed remote sensing image to obtain a current map patch of the target ground object type; The processing module is also used for performing grid processing on the current map patch to obtain a first grid corresponding to the current map patch, and performing grid processing on the historical map patch to obtain a second grid corresponding to the historical map patch; The determination module is used for determining a map patch change area of the target ground object type between the to-be-processed remote sensing image and the historical remote sensing image according to the first grid and the second grid; The processing module is also used for: Calculating the connected area of the connected grid in the first grid; According to the connected area of the connected grid in the first grid, the first grid is filtered to obtain a filtered first grid; The determination module is specifically used for: According to the filtered first grid and the second grid, the map patch change area is determined.

8. An electronic device, comprising: Comprise: A processor, a memory and a bus, the memory stores machine readable instructions executable by the processor, when the electronic device is running, the processor and the memory communicate through the bus, the processor executes the machine readable instructions to execute the determination method of the map patch change area in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to execute the determination method of the map patch change area in any one of claims 1-6.

Images