Method for constructing house risk portrait based on multi-source heterogeneous image

Abstract

The invention relates to the technical field of image processing, and discloses a method for constructing a house risk portrait based on a multi-source heterogeneous image, and the method comprises the steps: matching a simulation node set with a reference node set according to a simulation topological graph of each simulation graph in a house distribution graph set and a basic topological graph of a three-dimensional base graph; the house distribution graph set comprises a plurality of house risk simulation graphs stored in different forms; converting the successfully matched simulation nodes into a plane projection coordinate system of the three-dimensional base map; and semantically analyzing the simulation risk parameter set to determine a risk visualization elevation value of the successfully matched simulation node, and registering the simulation risk parameters to the three-dimensional base map to form a three-dimensional risk portrait taking the house as a basic unit. According to the method for constructing the house risk portrait based on the multi-source heterogeneous image, simulation images based on different coordinate references are registered to the same three-dimensional base map, and the three-dimensional risk portrait with a house as a basic unit is constructed.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method for constructing a building risk profile based on multi-source heterogeneous images. Background Technology

[0002] With the increasing demand for refined urban management, a large amount of housing risk simulation data has been accumulated in the daily supervision and risk assessment of departments such as housing construction, fire protection, and emergency response. This data originates from different business systems and is stored in various forms such as vector layers, raster layers, and grid color blocks, forming a typical multi-source heterogeneous image set. However, when processing the fusion mapping between these multi-source heterogeneous images and 3D base maps, it was found that simulation images from different sources are often based on different coordinate reference systems, such as local independent coordinate systems, global latitude and longitude coordinate systems, or grid coordinate systems. The lack of a unified transformation framework between these coordinate references leads to a significant spatial offset when images are superimposed, making direct registration to the same 3D base map difficult. For example, the conversion between local coordinate systems and standard geographic coordinate systems often fails due to missing parameters, resulting in misaligned building locations. Summary of the Invention

[0003] Therefore, the purpose of this invention is to overcome the problem of multi-base point nonlinear registration and fusion mapping between multi-source heterogeneous simulation images and three-dimensional base maps, and to provide a method for constructing a building risk profile based on multi-source heterogeneous images. This method registers simulation images based on different coordinate bases to the same three-dimensional base map, such as a three-dimensional city base map, to construct a three-dimensional risk profile with buildings as the basic unit.

[0004] To address the aforementioned technical problems, this invention provides a method for constructing a building risk profile based on multi-source heterogeneous images, comprising:

[0005] After transforming the planar coordinate system of each simulation map in the house distribution map set to the first planar projected coordinate system, the simulation topology map of each simulation map is extracted; the house distribution map set includes multiple house risk simulation maps stored in different formats, namely vector layer storage format, raster layer storage format, or grid color block storage format; the simulation topology map includes simulation node set, simulation edge set, and simulation risk parameter set;

[0006] Based on the simulation topology diagram and the basic topology diagram of the 3D base map, match the simulation node set with the baseline node set;

[0007] Transform the successfully matched simulation nodes into the second plane projection coordinate system of the 3D base map;

[0008] Semantic parsing of the simulation risk parameter set is used to determine the risk visualization elevation value of successfully matched simulation nodes, and the simulation risk parameters are registered onto the 3D base map to form a 3D risk profile with buildings as the basic unit.

[0009] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:

[0010] The method for constructing a building risk profile based on multi-source heterogeneous images described in this invention uses a technical approach of extracting a simulation topology map, matching simulation nodes with reference nodes, coordinate transformation, semantic resolution of elevation, and risk parameter registration to integrate three types of heterogeneous data—vector layers, raster layers, and grid color blocks—onto a three-dimensional base map, thereby constructing a three-dimensional risk profile with buildings as the basic unit.

[0011] Specifically, firstly, by extracting the simulation topology of each simulation map, the structural problem of inconsistent processing of data in different formats is addressed. Secondly, spatial correspondence is established by matching simulation nodes with reference nodes in the 3D base map. Based on this, the elevation values ​​of simulation nodes in the control point pairs in 3D space are determined through semantic parsing of simulation risk parameters, realizing the transformation from 2D planar risk information to 3D spatial positioning, allowing risks to be anchored to corresponding parts of the building. Finally, all risk parameters are registered to the building entities in the 3D base map, making each building a convergence point for multi-source risk information, forming a 3D risk profile with buildings as the basic unit. Attached Figure Description

[0012] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein:

[0013] Figure 1 This is a schematic flowchart of a method for constructing a building risk profile based on multi-source heterogeneous images in an embodiment of the present invention.

[0014] Figure 2 This is a schematic diagram of a rectangular block in an embodiment of the present invention. Detailed Implementation

[0015] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0016] The anchor points used in various risk images have fundamentally different semantic meanings. Vector layers rely on the inflection points of building property boundaries, raster layers use the center point of the raster cell as the reference, and grid color blocks use grid vertices or center points. This semantic inconsistency makes it difficult to establish accurate correspondences between key control points across data sources. For example, there is a logical conflict in matching building patch vertices with land parcel boundary points. Furthermore, simulation images from different sources are often based on different coordinate reference systems, such as local independent coordinate systems, global latitude and longitude coordinate systems, or grid coordinate systems. The lack of a unified transformation framework between these coordinate references leads to significant spatial offsets when images are overlaid, making direct registration to the same 3D base map difficult. In addition, urban areas exhibit complex nonlinear local distortion characteristics. Affected by factors such as terrain undulations and differences in building density, the deformation patterns vary in different areas. Traditional affine transformation methods can only handle global linear deformation and cannot adapt to local nonlinear changes, resulting in a significant reduction in registration accuracy in key areas. Therefore, this application proposes a method for constructing building risk profiles based on multi-source heterogeneous images.

[0017] Example 1: This example introduces a method for constructing a building risk profile based on multi-source heterogeneous images.

[0018] refer to Figure 1 The method for constructing a building risk profile based on multi-source heterogeneous images in this embodiment includes steps SS1 to SS4.

[0019] Step SS1: Match the simulation node set with the reference node set based on the simulation topology map of each simulation map in the house distribution map set and the basic topology map of the 3D base map.

[0020] In application, after transforming the planar coordinate system of each simulation map in the housing distribution map set to the first planar projected coordinate system, the simulation topology map of each simulation map is extracted. The housing distribution map set includes multiple housing risk simulation maps stored in different formats, namely vector layer storage format, raster layer storage format, or grid color block storage format. For example, the distribution map set includes housing risk simulation maps stored in vector layer format, housing risk simulation maps stored in raster layer format, and / or housing risk simulation maps stored in grid color block format.

[0021] Furthermore, vector layers typically represent geographic entities using geometric elements such as points, lines, and polygons, and can describe the boundaries and locations of buildings. Raster layers represent spatial data using regular raster cells. Grid color blocks are often used to represent the attribute distribution of irregular areas, such as areas divided by an irregular triangular mesh.

[0022] The difference between the first-plane projection coordinate system and the base coordinate system of the 3D base map is that the Z-axis of the 3D base map's plane projection coordinate system is 0. We can first identify the coordinate reference type of each initial simulation image; based on the transformation parameters between the initial simulation image's coordinate reference type and the 3D base map's plane projection coordinate system, we can use a seven-parameter transformation model or a four-parameter transformation model to batch transform the coordinates of all nodes in the simulation image to the first-plane projection coordinate system, thus transforming the simulation image's plane coordinate system to the first-plane projection coordinate system. This will not be elaborated further here.

[0023] In practical applications, the simulation topology map includes a set of simulation nodes, a set of simulation edges, and a set of simulation risk parameters. The set of simulation nodes includes spatial feature points in the building risk simulation map. The set of simulation edges includes the topological connections between simulation nodes; further, these connections include the straight-line distances and azimuth angles in radians between the simulation nodes. The set of simulation parameters includes the simulation risk parameter values ​​associated with each simulation node.

[0024] In actual implementation, the basic topology map includes a set of reference nodes. Furthermore, the basic topology map includes a set of reference nodes, a set of reference edges, and a set of elevations.

[0025] The reference node set includes land parcel boundary points, house number location points, remote sensing feature points, and building outline vertices. Specifically, land parcel boundary points can be extracted from cadastral maps; house number location points can be extracted from address databases; remote sensing feature points can be extracted from remote sensing imagery; and building outline vertices can be extracted from open-source map data.

[0026] The reference edge set includes edges formed by the spatial relationships between the reference nodes. Further, the spatial relationships include the straight-line distance, elevation difference, and azimuth angle in radians between the reference nodes. Even further, reference nodes belonging to the same land parcel or the same building outline are established as related edges, and reference adjacent edges are established between reference nodes whose spatial distance is less than the adjacency threshold.

[0027] The elevation set includes ground elevation values, building height, number of building floors, building floor height, and basement depth. In some embodiments, step SS1 includes steps SS11 to SS13.

[0028] Step SS11: Extract the first simulation topology map of the house risk simulation map stored in vector layer form.

[0029] In application, the simulation node set of the first simulated topology graph includes vertex nodes and center point nodes of building patches. The simulation edge set of the first simulated topology graph includes boundary edges of patches, lines connecting the vertices of the center points of patches, and adjacent edges of patches. The simulation risk parameter set of the first simulated topology graph includes building code, building age, structural type, and safety assessment level. Among them, the safety assessment level is A, B, C, or D.

[0030] In practical applications, the method for extracting the first simulation topology map in step SS11 includes steps SS111 to SS116.

[0031] Step SS111: Traverse the polygonal boundary of each house patch and obtain the coordinates of all vertices of each polygonal boundary as the vertex nodes of the house patch.

[0032] In application, the vertex nodes of a house feature form the basis of its shape. For example, a rectangular house feature has four vertex nodes, while an L-shaped house feature has at least six vertex nodes.

[0033] In practical applications, vertex coordinates are planar coordinates, and the unit can be meters.

[0034] Step SS112: Calculate the geometric center point of each house patch based on the vertex nodes of the house patch, and use the geometric center point of the house patch as the center point node of the house patch.

[0035] In application, the X-coordinate of the geometric center point is equal to the average of the X-coordinates of all vertices, and the Y-coordinate of the geometric center point is equal to the average of the Y-coordinates of all vertices. Furthermore, the center point node of a house patch is used to represent the overall position of the house patch.

[0036] Step SS113: Establish boundary edges between adjacent vertex nodes of each house patch and record the edge length and azimuth.

[0037] When applied, the side length is the distance between two points, and the azimuth angle is the azimuth angle from the first vertex node to the second vertex node, in radians.

[0038] In practical applications, the boundary edges of a map feature can describe the external outline of a building map feature.

[0039] Step SS114: Establish a line connecting the center point and vertex of each house patch to the center point of each vertex node, and record the line length and azimuth.

[0040] When applied, the line connecting the vertices of the center point of the map patch can reflect the internal geometric features of the house map patch.

[0041] Step SS115: Establish adjacent edges between the center point nodes of adjacent house patches, and record the adjacent distance and azimuth.

[0042] When applied, adjacent edges of a feature can capture the spatial relationships between building features.

[0043] In practical applications, if two townhouses share a wall, their house patches are considered adjacent. Alternatively, if two townhouses do not share a boundary, but the distance between their center points is less than the adjacency threshold, then their house patches are considered adjacent.

[0044] In practice, the adjacency threshold is determined based on the node distribution density. The adjacency threshold is half the average distance between all node pairs.

[0045] Step SS116: Use the attribute fields of each house patch as the simulation risk parameter set for the corresponding center point node.

[0046] Step SS12: Extract the second simulation topology map of the house risk simulation map stored in the form of a raster layer.

[0047] In application, the simulation node set of the second simulation topology graph includes the center point nodes of the grid cells. The simulation edge set of the second simulation topology graph includes adjacent grid edges. The simulation risk parameter set of the second simulation topology graph includes the fire risk index. Furthermore, the fire risk index ranges from 0 to 100.

[0048] In practical applications, the method for extracting the second simulation topology map in step SS12 includes steps SS121 to SS123.

[0049] Step SS121: Traverse each grid cell, calculate the coordinates of the geometric center point of each grid cell based on the grid row and column number and spatial resolution, and use the coordinates of the center point of the grid cell as the grid cell center point node.

[0050] When applying this method, set the geographic coordinates of the origin of the raster layer to be... The X-axis resolution of the grid cell is ΔX, and the Y-axis resolution of the grid cell is ΔY.

[0051] In practical applications, the geometric center coordinates of the grid cell with row number r and column number z are: .

[0052] Step SS122: Establish grid adjacency edges between the center point nodes of adjacent grid cells and record the adjacency distance and azimuth angle.

[0053] In application, each grid cell center node establishes a grid adjacency edge with the four adjacent grid cell center nodes that are directly above, below, to the left, and to the right of the grid cell.

[0054] Step SS123: Use the attribute fields of each grid cell as the simulation risk parameter set for the corresponding center point node.

[0055] Step SS13: Extract the third simulation topology map of the house risk simulation map stored in the form of grid color blocks.

[0056] In application, the simulation node set of the third simulation topology graph includes grid cell vertex nodes and grid cell center point nodes. The simulation edge set of the third simulation topology graph includes grid boundary edges, lines connecting grid center point vertices, and grid adjacent edges. The simulation risk parameter set of the third simulation topology graph includes accident risk levels. Furthermore, the accident risk levels are high risk, medium risk, or low risk.

[0057] In practical applications, the method for extracting the third simulation topology map in step SS13 includes steps SS131 to SS136.

[0058] Step SS131: Traverse the polygon boundary of each grid cell and obtain the coordinates of all vertices of each polygon boundary as the vertex nodes of the grid cell.

[0059] In application, edge detection algorithms from image processing can be used to identify grid boundaries, and then contour extraction algorithms can be used to track and extract the vertex coordinates of each grid cell, which will not be elaborated here.

[0060] Step SS132: Calculate the geometric center point of each mesh cell based on the vertex nodes of the mesh cells, and use the geometric center point of the mesh cell as the center point node of the mesh cell.

[0061] In application, the center point node of the grid cell is used to determine the spatial location of each grid cell.

[0062] Step SS133: Establish grid boundary edges between adjacent vertex nodes of each grid cell and record the edge length and azimuth.

[0063] Step SS134: Establish a line connecting the center point of each grid cell to each vertex node, and record the line length and azimuth.

[0064] In application, the connection between the vertices of the grid center point is used to construct the topological connections within the grid cell.

[0065] Step SS135: Establish mesh adjacency edges between the center point nodes of adjacent mesh cells and record the adjacency distance and azimuth angle.

[0066] When applying this feature, you can check if two grid cells have at least two identical consecutive vertex nodes. If so, the two grid cells are adjacent.

[0067] Step SS136: Use the attribute fields of each mesh cell as the simulation risk parameter set for the corresponding center point node.

[0068] In some embodiments, step SS1 further includes step SS14. Further, step SS1 further includes step SS15.

[0069] Step SS14: Perform coordinate matching between the simulation nodes and the reference nodes.

[0070] When applied, step SS14 includes steps SS141 to SS142.

[0071] Step SS141: Calculate the distance between the simulation node and the reference node, and select the reference node with the smallest distance from the simulation node as the first candidate node for the simulation node.

[0072] In application, the distance between the simulation node and the reference node can be Euclidean distance.

[0073] Step SS142: When the distance between the simulation node and the corresponding first candidate node is less than a preset threshold, the simulation node and the corresponding first candidate node are paired as a control point pair.

[0074] When applying the simulation, the preset threshold is determined based on the scale of the simulation map: when the scale is greater than or equal to 1:500, the preset threshold is 2 meters; otherwise, the preset threshold is 5 meters.

[0075] In practical applications, for the first simulated topology map: the vertex nodes of building patches are matched with the land parcel boundary points and building outline vertices, and the center point nodes of building patches are matched with the address locations. For the second simulated topology map: the center point nodes of grid cells are matched with remote sensing feature points. For the third simulated topology map: the vertex nodes of grid cells are matched with the land parcel boundary points and building outline vertices, and the center point nodes of grid cells are matched with the address locations.

[0076] Step SS15: When the distance between the simulation node and the corresponding first candidate node is greater than or equal to a preset threshold, the simulation node is marked as a node to be matched, and the node to be matched is subjected to topology matching: when the topology matching is successful, the node to be matched is paired with the corresponding reference node as a control point pair; otherwise, the node to be matched is marked as an isolated node.

[0077] In order to overcome the limitations of coordinate matching and improve the applicability of working conditions with local deformation or large coordinate deviation, topological matching is performed on the nodes to be matched.

[0078] In practical applications, once a simulation node is marked as an isolated node, the corresponding simulation node will not participate in subsequent steps of risk profile construction.

[0079] In actual implementation, the rules for matching the topology structure include three cases.

[0080] For the first simulation topology graph, the topology matching includes: step SS151.

[0081] Step SS151: Obtain the neighboring nodes of the node to be matched, and calculate the cost function value using the reference node that is paired with the neighboring node as a control point pair and the first candidate node of the node to be matched: when the cost function value is less than the preset threshold, pair the node to be matched and its first candidate node as a control point pair.

[0082] When applied, the cost function value includes the following formula:

[0083] ;

[0084] In the formula, This represents the total number of control point pairs that have already been paired with all the neighboring nodes of the node to be matched; The numbers of the adjacent nodes that have been paired as control point pairs. It is a positive integer, and ; and The coordinates of the first candidate node to be matched; and To be with the number The adjacent nodes are paired as the coordinates of the reference node of the control point pair; For the node to be matched and the node with the number The distance between adjacent nodes; For the node to be matched and the node with the number The azimuth angle between adjacent nodes.

[0085] For the second simulation topology graph, the topology matching includes steps SS152 to SS153.

[0086] Step SS152: Based on the grid cell resolution, match the position of the adjacent node to the east of the center point node of the matched grid cell at a location ΔX east of the current matching reference node, and form a corresponding control point pair. Here, ΔX is the X-axis resolution of the grid cell.

[0087] When applying the algorithm, the center point node of the matched grid cell is denoted as... ;and The matching reference node is ; The eastern adjacent node is ; East of ΔX is ;Will and The match is a pair of control points.

[0088] Step SS153: Match the position of the adjacent node to the north of the center point node of the matched grid cell at a location ΔY north of the current matching reference node, and form a corresponding control point pair. Here, ΔY is the Y-axis resolution of the grid cell.

[0089] When applying the algorithm, the center point node of the matched grid cell is denoted as... ;and The matching reference node is ; The adjacent node on the north side is ; North of ΔY ;Will and The match is a pair of control points.

[0090] For the third simulation topology graph, the topology matching includes: step SS154.

[0091] Step SS154: When three vertex nodes in the mesh cell to which the node to be matched belongs have already matched control point pairs, determine the matching position of the node to be matched by the parallelogram rule and pair them as control point pairs; otherwise, mark the node to be matched as an isolated node.

[0092] In application, the mesh element includes a first node, a second node, a third node, and a fourth node. When the fourth node is a node to be matched, but the first, second, and third nodes have already been matched with control point pairs, based on the properties of a parallelogram, the coordinates of the topological candidate node are calculated using the first candidate nodes of the first, second, and third nodes, and the topological candidate node is paired with the fourth node as a control point pair.

[0093] Step SS2: Transform the successfully matched simulation nodes into the planar projection coordinate system of the 3D base map.

[0094] In application, the coordinates of the simulation nodes of each control point pair in the simulation topology map are transformed to the second-plane projection coordinate system of the 3D base map to obtain the third-plane projection coordinates. The basic coordinate system of the 3D base map includes the second-plane projection coordinate system and the elevation coordinate axis. The second-plane projection coordinate system is the projection of the 3D base map's basic coordinate system onto a two-dimensional plane, with its Z-axis coordinate being 0, used to describe the spatial position of features in the horizontal direction; the elevation coordinate axis is used to describe the height of features in the vertical direction.

[0095] In practical applications, step SS2 includes steps SS21 to SS25.

[0096] Step SS21: Divide the space of the simulation topology map into multiple rectangular blocks, each rectangular block containing at least 3 control point pairs.

[0097] In application, the side length of each rectangular block is determined based on the characteristic scale of spatial deformation. When there are fewer than three control point pairs within a rectangular block, the block is merged with adjacent rectangular blocks. For example, in urban centers with dense terrain features, high control point pair density, and a large rate of change in spatial deformation, the side length of the rectangular block can be 200 meters; in suburban areas with sparse terrain features, low control point pair density, and a small rate of change in spatial deformation, the side length of the rectangular block can be 1000 meters; the side length of the rectangular blocks in both urban centers and suburban areas is 500 meters.

[0098] Step SS22: Determine the affine transformation formula for the center point of the rectangular block using the least squares method and the control point pairs within each rectangular block.

[0099] When applied, the affine transformation formula for the center point of the rectangular block includes:

[0100] ;

[0101] ;

[0102] In the formula, and Transformed coordinates of the center point of the rectangular block; and These are the coordinates of the center point of the rectangular block; , , , , , The affine transformation parameters, which are all affine transformation formulas for the center points of rectangular blocks, can be determined by least squares fitting. In some embodiments, It can be between 0.90 and 1.10; It can range from -0.20 to 0.20; It can range from -200 to 200; It can range from -0.20 to 0.20; It can be between 0.90 and 1.10; It can range from -200 to 200.

[0103] In practical applications, the affine transformation formula for the center point of a rectangular block can be a general affine transformation formula for rectangular blocks.

[0104] Step SS23: For the center point of each rectangular block, the affine transformation parameters of the center point of the rectangular block and the affine transformation parameters of the center points of the first group of adjacent rectangular blocks of the rectangular block are weighted and averaged to obtain the smoothed affine transformation formula of the rectangular block.

[0105] In application, the first group of adjacent rectangular blocks includes the four adjacent rectangular blocks located above, below, to the left, and to the right of the rectangular block. Further, refer to... Figure 2 The four adjacent rectangular blocks above, below, left, and right of the rectangular block are the first rectangular block, the second rectangular block, the third rectangular block, and the fourth rectangular block.

[0106] In practical applications, the weights of the weighted average are determined based on the spatial relationship between the rectangular block and its adjacent rectangular blocks.

[0107] In some embodiments, the weight of a rectangular block is 0.5; the weight of adjacent rectangular blocks is 0.125. Specifically: ;

[0108] In the formula, The parameters are the affine transformation parameters after smoothing the rectangular block; The affine transformation parameters for the rectangular block; The parameters are the affine transformation parameters for the first rectangular block; Let the affine transformation parameters of the rectangular block be denoted as ; The parameters for the affine transformation of the third rectangular block; The parameters are the affine transformation parameters for the fourth rectangular block.

[0109] Step SS24: Based on the center points of the second group of adjacent rectangular blocks of the rectangular block, calculate the relative position of the simulation node of any control point pair in the rectangular block within the first region formed by the center points of the second group of adjacent rectangular blocks.

[0110] In application, the second group of adjacent rectangular blocks includes the four adjacent rectangular blocks located at the top left, top right, bottom left, and bottom right of the original rectangular block. Further, refer to... Figure 2 The four adjacent rectangular blocks at the top left, top right, bottom left, and bottom right are the fifth, sixth, seventh, and eighth rectangular blocks, respectively.

[0111] In practical applications, the relative positions of the simulation nodes within the first region include:

[0112] ;

[0113] ;

[0114] In the formula, The relative X-axis position of the simulation node within the first region; The relative Y-axis position of the simulation node within the first region; and These are the initial coordinates of the simulation nodes; and The coordinates are the center points of the adjacent rectangular blocks in the upper left corner of the rectangular block; and The coordinates are the center points of the adjacent rectangular blocks to the upper right of the rectangular block; and The coordinates are the center points of the adjacent rectangular blocks to the lower left of the rectangular block.

[0115] Step SS25: Based on the smoothed affine transformation formula, according to the relative position of the simulation node in the first region, perform bilinear interpolation on each parameter of the smoothed affine transformation formula of the rectangular block to obtain the transformed coordinates of the simulation node.

[0116] In application, each parameter of the smoothed affine transformation formula for the rectangular block is bilinearly interpolated, including the following formula:

[0117] ;

[0118] ;

[0119] ;

[0120] In the formula, For any smoothed affine transformation parameter of the simulation node, for example, for , , , , or ; The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks in the upper left corner of the rectangular block; The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks to the upper right of the rectangular block. The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks at the lower left of the rectangular block; The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks to the lower right of the rectangular block. This is the interpolation result for the upper boundary; This is the interpolation result for the lower boundary.

[0121] In practical applications, the transformed coordinates of the simulation node include:

[0122] ;

[0123] ;

[0124] In the formula, and The transformed coordinates of the simulation nodes; and These are the coordinates of the simulation nodes; , , , , , All parameters are affine transformation parameters after bilinear interpolation of the simulation nodes.

[0125] Step SS3: Semantically analyze the simulation risk parameter set to determine the risk visualization elevation value of the successfully matched simulation nodes, and register the simulation risk parameters onto the 3D base map to form a 3D risk profile with houses as the basic unit.

[0126] In application, the simulation risk parameter set is semantically parsed to determine the risk visualization elevation value of the simulation node in the control point pair, and the simulation risk parameters are registered to the building entities on the 3D base map to form a 3D risk profile with the building as the basic unit.

[0127] In practical applications, first determine whether the control point, after being transformed into the planar projection coordinate system of the 3D base map, falls within the outline of the building entity: if not, the risk visualization elevation value = 0. If it falls within the outline of the building entity, the risk visualization elevation value is determined based on the simulation risk parameter set.

[0128] For the first simulation topology map: read the safety assessment level from the simulation risk parameter set, and determine the visualized elevation value of the structural safety risk based on the safety assessment level.

[0129] When the safety assessment level is the highest, such as Level D, the visualized elevation value of structural safety risk = ground elevation + building height. When the safety assessment level is just below the highest, such as Level C, the visualized elevation value of structural safety risk = ground elevation + 1 / 2 building height. Otherwise, the visualized elevation value of structural safety risk = 0.

[0130] For the second simulation topology map: read the fire risk index from the simulation risk parameter set, and determine the visualized elevation value of the fire risk based on the label information of the fire risk index.

[0131] If the label information includes a floor number, then the visualized elevation value of fire risk = ground elevation + (floor number - 1) * building height; the unit of 1 is the same as the unit of the floor number. If the label information includes a roof, then the visualized elevation value of fire risk = ground elevation + building height + 1; the unit of 1 is the same as the unit of the building height. If the label information includes a basement or garage, then the visualized elevation value of fire risk = ground elevation - basement depth. Otherwise, the visualized elevation value of fire risk = ground elevation.

[0132] For the third simulation topology map, the accident risk level is read from the simulation risk parameter set. When the accident risk level is not empty, the visualized elevation value of production safety risk = the ground elevation value.

[0133] In practical implementation, the visualized elevation value determined through semantic parsing is used as the Z-axis coordinate in the basic coordinate system of the 3D base map. This Z-axis coordinate, together with the projection coordinates of the third plane, constitutes the 3D spatial coordinates of the control point centering simulation node. These 3D spatial coordinates are then used as the coordinates of the control point centering simulation node on the 3D base map, thereby achieving the registration of simulation risk parameters onto the 3D base map and forming a 3D risk profile with buildings as the basic unit. Those skilled in the art should understand that the embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0134] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0135] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0136] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0137] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for constructing a housing risk profile based on multi-source heterogeneous images, characterized in that, include: Based on the simulation topology map and the basic topology map of the 3D base map of each simulation map in the house distribution map set, match the simulation node set with the reference node set; The building distribution map atlas includes multiple building risk simulation maps stored in different formats, such as vector layer storage, raster layer storage, or grid color block storage; the simulation topology map includes a simulation node set, a simulation edge set, and a simulation risk parameter set. Transform the successfully matched simulation nodes into the planar projection coordinate system of the 3D base map; Semantic parsing of the simulation risk parameter set is used to determine the risk visualization elevation value of successfully matched simulation nodes, and the simulation risk parameters are registered onto the 3D base map to form a 3D risk profile with buildings as the basic unit.

2. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 1, characterized in that, The process of matching the simulation node set with the baseline node set based on the simulation topology graph and the base topology graph includes: The coordinate matching between the simulation node and the reference node includes: calculating a first distance between the simulation node and the reference node, and taking the reference node with the smallest first distance as the first candidate node of the simulation node; when the second distance between the simulation node and the corresponding first candidate node is less than a preset threshold, the simulation node and the corresponding first candidate node are paired as a control point pair.

3. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 2, characterized in that, The coordinate matching between the simulation node and the reference node also includes: For the first simulation topology of the house risk simulation map stored in the form of a vector layer: match the vertex nodes of the house patch with the land boundary points and the building outline vertices, and match the center point nodes of the house patch with the address points; For the second simulation topology map of the building risk simulation map stored in the form of a raster layer: match the center point node of the raster cell with the remote sensing ground feature point; For the third simulation topology of the house risk simulation map stored in the form of grid color blocks: match the grid cell vertex nodes with the land boundary points and building outline vertices, and match the grid cell center point nodes with the house number positioning points.

4. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 2, characterized in that, The step of matching the simulation node set with the baseline node set based on the simulation topology graph and the base topology graph also includes: When the second distance is greater than or equal to a preset threshold, the simulation node is marked as a node to be matched, and the node to be matched is subjected to topology matching: when the topology matching is successful, the node to be matched is paired with the corresponding reference node as a control point pair; otherwise, the node to be matched is marked as an isolated node.

5. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 4, characterized in that, For the first simulation topology map of the house risk simulation map stored in the form of a vector layer, the topology matching includes: obtaining the neighboring nodes of the node to be matched, and calculating the cost function value using the reference node that is paired with the neighboring node as a control point pair and the first candidate node of the node to be matched; when the cost function value is less than a preset threshold, the node to be matched and its first candidate node are paired as a control point pair. For the second simulation topology map of the building risk simulation map stored in the form of a raster layer, the topology matching includes: based on the resolution of the raster cells, matching the position of the eastern adjacent node of the center point node of the matched raster cell at a location ΔX east of the current matching reference node, and forming a corresponding control point pair; matching the position of the northern adjacent node of the center point node of the matched raster cell at a location ΔY north of the current matching reference node, and forming a corresponding control point pair; wherein, ΔX is the X-axis resolution of the raster cell, and ΔY is the Y-axis resolution of the raster cell; For the third simulation topology map of the house risk simulation map stored in the form of grid color blocks, the topology matching includes: when three vertex nodes in the grid cell to which the node to be matched belongs have already matched control point pairs, the matching position of the node to be matched is determined by the parallelogram rule and paired as a control point pair; otherwise, the node to be matched is marked as an isolated node.

6. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 5, characterized in that, The cost function value includes the following formula: ； In the formula, This represents the total number of control point pairs that have already been paired with all the neighboring nodes of the node to be matched; The numbers of the adjacent nodes that have been paired as control point pairs. It is a positive integer, and ; and The coordinates of the first candidate node to be matched; and To be with the number The adjacent nodes are paired as the coordinates of the reference node of the control point pair; For the node to be matched and the node with the number The distance between adjacent nodes; For the node to be matched and the node with the number The azimuth angle between adjacent nodes.

7. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 1, characterized in that, The process of transforming the successfully matched simulation nodes into the planar projection coordinate system of the 3D base map includes: The space of the simulation topology map is divided into multiple rectangular blocks, and each rectangular block contains at least 3 pairs of control points; The affine transformation formula for determining the center point of the rectangular block is obtained by using the least squares method and the control points within each rectangular block. For each rectangular block center point, the affine transformation parameters of the rectangular block center point and the affine transformation parameters of the center points of the first group of adjacent rectangular blocks are weighted and averaged to obtain the smoothed affine transformation formula of the rectangular block; wherein, the first group of adjacent rectangular blocks includes the four adjacent rectangular blocks above, below, left and right of the rectangular block. Based on the center point of the second group of adjacent rectangular blocks of the rectangular block, calculate the relative position of the simulation node of any control point pair in the rectangular block within the first region formed by the center points of the second group of adjacent rectangular blocks; wherein, the second group of adjacent rectangular blocks includes four adjacent rectangular blocks: the upper left, upper right, lower left, and lower right of the rectangular block. Based on the smoothed affine transformation formula, bilinear interpolation is performed on each parameter of the smoothed affine transformation formula of the rectangular block according to the relative position of the simulation node in the first region to obtain the transformed coordinates of the simulation node.

8. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 7, characterized in that, The relative positions of the simulation nodes in the first region include: ； ； In the formula, The relative X-axis position of the simulation node within the first region; The relative Y-axis position of the simulation node within the first region; and These are the coordinates of the simulation nodes; and The coordinates are the center points of the adjacent rectangular blocks in the upper left corner of the rectangular block; and The coordinates are the center points of the adjacent rectangular blocks to the upper right of the rectangular block; and The coordinates are the center points of the adjacent rectangular blocks to the lower left of the rectangular block; Each parameter of the smoothed affine transformation formula for the rectangular block is subjected to bilinear interpolation, including: ； ； ； In the formula, For any smoothed affine transformation parameter of the simulation node; The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks in the upper left corner of the rectangular block; The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks to the upper right of the rectangular block. The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks at the lower left of the rectangular block; The smoothed affine transformation parameters are the center points of the adjacent rectangular blocks to the lower right of the rectangular block. This is the interpolation result for the upper boundary; This is the interpolation result for the lower boundary; The transformed coordinates of the simulation node include: ； ； In the formula, and The transformed coordinates of the simulation nodes; , , , , , All parameters are affine transformation parameters after bilinear interpolation of the simulation nodes.

9. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 1, characterized in that, The semantic parsing simulation risk parameter set is used to determine the risk visualization elevation value of successfully matched simulation nodes, including: For the first simulation topology of the building risk simulation map stored in vector layer form: read the safety assessment level from the simulation risk parameter set: when the safety assessment level is the highest level, the visualized elevation value of structural safety risk = ground elevation value + building height; when the safety assessment level is second only to the highest level, the visualized elevation value of structural safety risk = ground elevation value + 1 / 2 building height; otherwise, the visualized elevation value of structural safety risk = 0. For the second simulation topology map of the building risk simulation map stored in the form of a raster layer: read the fire risk index from the simulation risk parameter set, and determine the visual elevation value of the fire risk based on the label information of the fire risk index; For the third simulation topology map of the building risk simulation map stored in the form of grid color blocks: read the accident risk level from the simulation risk parameter set. When the accident risk level is not empty, the visualized elevation value of production safety risk = the ground elevation value.

10. The method for constructing a building risk profile based on multi-source heterogeneous images according to claim 9, characterized in that, The visualized elevation values ​​for fire risk include: If the label information includes a floor number, then the visualized elevation value of fire risk = ground elevation value + (floor number - 1)·building floor height; If the label information includes a roof, then the visualized elevation value of the fire risk = ground elevation value + building height + 1; If the label information includes underground or garage, then the visualized elevation value of fire risk = surface elevation value - basement depth; Otherwise, the visualized elevation value of fire risk equals the ground elevation value.

Images