An indoor and outdoor topology connection method and system based on door hole and wall relationship

Abstract

The application provides an indoor and outdoor topology connection method and system based on the relationship between doorways and walls, relates to the technical field of digital model construction, and through construction of an explicit topology connection graph taking doorways as connection edges and indoor and outdoor space units as nodes, makes the registration result upgrade from simple coordinate transformation to structured topology data that can directly support indoor and outdoor integrated navigation and emergency evacuation simulation, can automatically complete indoor and outdoor topology connection determination, does not need manual pre-calibration of anchor points, and is expected to support batch automation processing under the city-level existing building survey scene through theoretical analysis, and solves the globally optimal matching by combining wall continuity, geometric similarity and orientation consistency three clues to fuse and score, compared with the matching mode that only depends on single geometric similarity, greatly improves the matching rate, and thus gives consideration to automation efficiency and batch reliability.

Description

Technical Field

[0001] This application relates to the field of digital model construction, specifically to an indoor and outdoor topological connection method and system based on the relationship between doorways and walls. Background Technology

[0002] Current 3D building reconstruction workflows typically implement outdoor oblique photography / airborne LiDAR reconstruction and indoor mobile laser scanning / panoramic reconstruction as two separate technical lines. However, due to the scarcity of visually overlapping indoor and outdoor data and significant differences in acquisition lighting conditions, the motion-reconstructed structural reconstruction results on both indoor and outdoor sides are usually independent, with no natural geometric or topological relationship between them.

[0003] Some studies have begun to combine the two technical lines, but the robustness of this type of method decreases when the number of door and window candidates extracted from the indoor and outdoor sides is unequal (e.g., missed detections due to furniture obstruction indoors, and false detections due to external obstructions outdoors). Moreover, its algorithm output still takes point set registration transformation as the final goal, without further structuring the matching results into a spatial topology map carrying semantic information such as doorway attributes, orientation, and wall continuity. It also does not design a confidence feedback and rematch closed-loop mechanism for mismatches. Therefore, in the scenario of surveying existing buildings with many obstructions, the degree of automation and verifiability are insufficient.

[0004] Therefore, there is an urgent need for a topology matching method that can improve the matching rate of indoor and outdoor 3D model fusion. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a method and system for indoor and outdoor topological connection based on the relationship between doorways and walls.

[0006] According to one aspect of this application, an indoor-outdoor topological connection method based on the relationship between doorways and walls is provided, comprising: acquiring an outdoor 3D model and an indoor 3D model, and extracting components from the outdoor 3D model and the indoor 3D model to obtain an indoor-outdoor component set; identifying doorways in the indoor-outdoor component set to obtain a candidate doorway set; for each candidate doorway in the candidate doorway set, determining the attributes of the candidate doorway and filtering to obtain an entrance doorway with the attribute of an entrance opening; wherein, the attributes of the candidate doorway include an entrance opening and an open space opening; and based on the distance and position of the outdoor entrance doorway and the indoor entrance doorway... The system identifies the continuous wall between the outdoor entrance doorway and the indoor entrance doorway, along with its corresponding confidence level. Based on the outdoor and indoor entrance doorways corresponding to the continuous wall, a bipartite graph is constructed, with the outdoor entrance doorway as the left node and the indoor entrance doorway as the right node. Based on the geometric similarity of the doorways, their orientation consistency, and the continuity of the wall, the system solves for the globally optimal matching result of the indoor and outdoor doorways on the bipartite graph. Using outdoor and indoor space units as nodes and the doorways confirmed by the globally optimal matching result as connecting edges, a unified indoor and outdoor topology connection graph is constructed.

[0007] In one embodiment, extracting components from the outdoor 3D model and the indoor 3D model to obtain an indoor and outdoor component set includes: performing semantic segmentation on the outdoor 3D model to extract outdoor components; wherein the outdoor components include exterior wall panels, roof panels, and ground panels; parsing the component tree of the indoor 3D model to extract indoor components; wherein the indoor components include walls, doors, windows, and spaces; and storing the outdoor components and the indoor components hierarchically by type as a component attribute table.

[0008] In one embodiment, identifying door openings in the set of indoor and outdoor components to obtain a candidate door opening set includes: using a boundary hole detection algorithm to identify through holes on the exterior wall panel; performing minimum bounding rectangle fitting on the boundary of the through hole to obtain outdoor candidate door openings; if the interior 3D model has a door component, extracting the geometric bounding box of the door component as an indoor candidate door opening; otherwise, performing a hole detection algorithm on the wall to obtain indoor candidate door openings; and obtaining the candidate door opening set based on the outdoor candidate door openings and the indoor candidate door openings.

[0009] In one embodiment, determining the attributes of each candidate doorway in the candidate doorway set includes: for each candidate doorway in the candidate doorway set, taking the outer normal vector of the wall where the candidate doorway is located as the orientation vector of the candidate doorway, calculating the coordinates of the center point of the doorway, width, height, and height above the ground of the candidate doorway; performing a nearest neighbor radius search with the center point of the doorway as the center of the sphere, and statistically analyzing the wall density distribution within the neighborhood; and distinguishing the attributes of the candidate doorway based on the wall density distribution within the neighborhood.

[0010] In one embodiment, determining the continuous wall between the outdoor entrance doorway and the indoor entrance doorway and its corresponding confidence level based on the distance and location information of the outdoor entrance doorway and the indoor entrance doorway includes: calculating the Euclidean distance between the center points of the outdoor entrance doorway and the indoor entrance doorway on the horizontal projection plane; if the Euclidean distance is less than a preset distance threshold, calculating the cosine similarity of the angle between the normal vector of the outdoor wall surface corresponding to the outdoor entrance doorway and the normal vector of the indoor wall corresponding to the indoor entrance doorway; if the cosine similarity is less than a preset similarity threshold, ray sampling is performed on the spatial region between the outdoor wall surface and the indoor wall along the direction of the normal vector of the outdoor wall surface, and the proportion of sampling points falling within a reasonable wall thickness range is counted; if the proportion is not lower than a preset proportion threshold, the spatial region is determined to be a continuous wall, and the confidence level of the continuous wall is calculated.

[0011] In one embodiment, constructing a bipartite graph based on the outdoor entrance doorway and indoor entrance doorway corresponding to the continuous wall includes: retrieving the room unit adjacent to the indoor entrance doorway corresponding to the continuous wall; if the height of the room unit is within a preset height range, then constructing a bipartite graph based on the outdoor entrance doorway and indoor entrance doorway, and establishing a candidate edge for each pair of outdoor entrance doorways and indoor entrance doorways.

[0012] In one embodiment, the step of finding the globally optimal matching result of the inner and outer door openings on the bipartite graph based on the geometric similarity of the door openings, the consistency of orientation, and the continuity of the walls includes: calculating the geometric similarity, the consistency of orientation score, and the comprehensive weight of each candidate edge in the bipartite graph; using the Hungarian algorithm to find the maximum weight bipartite graph matching result on the bipartite graph based on the geometric similarity, the consistency of orientation score, and the comprehensive weight, and obtaining the matching result; if the comprehensive weight of each edge in the matching result is greater than a preset weight threshold, then the matching result is determined to be the globally optimal matching result.

[0013] In one embodiment, the step of constructing a unified indoor-outdoor topology connection graph using outdoor space units and indoor space units as nodes and doorways confirmed by the global optimal matching result as connecting edges includes: constructing a unified indoor-outdoor topology connection graph using outdoor space units and indoor space units as nodes and doorways confirmed by the global optimal matching result as connecting edges.

[0014] In one embodiment, the indoor-outdoor topology connection method based on the relationship between doorways and walls further includes: searching the unified indoor-outdoor topology connection graph for paths from each indoor room node to an outdoor node; if all indoor rooms have paths to outdoor nodes, then the final topology connection graph is obtained.

[0015] According to another aspect of this application, an indoor-outdoor topological connection system based on the relationship between doorways and walls is provided, comprising: a component set extraction module, used to acquire an outdoor 3D model and an indoor 3D model, and extract components from the outdoor 3D model and the indoor 3D model to obtain an indoor-outdoor component set; a candidate doorway identification module, used to identify doorways in the indoor-outdoor component set to obtain a candidate doorway set; an entrance doorway determination module, used to determine the attributes of each candidate doorway in the candidate doorway set and filter to obtain entrance doorways with the attribute of entrance opening; wherein, the attributes of the candidate doorways include entrance opening and open space opening; and a continuous wall determination module, used to determine the relationship between the outdoor entrance doorway and the indoor entrance doorway. The system uses the distance and location information of doorways to determine the continuous wall between the outdoor entrance doorway and the indoor entrance doorway, along with their corresponding confidence levels. A bipartite graph construction module is used to construct a bipartite graph based on the outdoor and indoor entrance doorways corresponding to the continuous wall; wherein the bipartite graph uses the outdoor entrance doorway as the left node and the indoor entrance doorway as the right node. An optimal matching solution module is used to solve for the globally optimal matching result of the indoor and outdoor doorways on the bipartite graph based on the geometric similarity of the doorways, orientation consistency, and wall continuity. A topology graph construction module is used to construct a unified indoor and outdoor topology connection graph using outdoor and indoor space units as nodes and the doorways confirmed by the globally optimal matching result as connecting edges.

[0016] This application provides a method and system for indoor and outdoor topological connection based on the relationship between doorways and walls. The method involves acquiring outdoor and indoor 3D models, extracting components from both models to obtain a set of indoor and outdoor components, identifying doorways within the component set to obtain a set of candidate doorways, determining the attributes of each candidate doorway and filtering for entrance doorways with the attribute of "entrance opening," where the attributes include "entrance opening" and "opening into space." Based on the distance and location information between the outdoor and indoor entrance doorways, the method identifies the continuous wall between them and its corresponding confidence level. A bipartite graph is constructed based on the outdoor and indoor entrance doorways corresponding to the continuous wall, where the outdoor entrance doorway is the left node and the indoor entrance doorway is the right node. The method further considers the geometric similarity and orientation of the doorways. Consistency and wall continuity are addressed by finding the globally optimal matching result for interior and exterior door openings on a bipartite graph. Using outdoor and indoor space units as nodes and door openings confirmed by the globally optimal matching result as connecting edges, a unified indoor and outdoor topological connection graph is constructed. By constructing an explicit topological connection graph with door openings as connecting edges and indoor and outdoor space units as nodes, the registration result is upgraded from a simple coordinate transformation to structured topological data that can directly support integrated indoor and outdoor navigation and emergency evacuation simulation. It can automatically complete the determination of indoor and outdoor topological connections without the need for manual pre-calibration of anchor points. Theoretical analysis suggests it can support batch automated processing in urban-level existing building survey scenarios. Furthermore, by combining wall continuity, geometric similarity, and orientation consistency as three criteria for fusion scoring to solve for the globally optimal match, the matching rate is significantly improved compared to matching methods that rely solely on geometric similarity, thus balancing automation efficiency and critical reliability. Attached Figure Description

[0017] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1 This is a flowchart illustrating an exemplary embodiment of the present application of an indoor-outdoor topological connection method based on the relationship between a doorway and a wall.

[0019] Figure 2 This is a schematic diagram of an indoor-outdoor topological connection system based on the relationship between a doorway and a wall, provided in an exemplary embodiment of this application.

[0020] Figure 3 This is a structural diagram of an electronic device provided in an exemplary embodiment of this application. Detailed Implementation

[0021] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0022] Figure 1 This is a flowchart illustrating an exemplary embodiment of the present application of an indoor-outdoor topological connection method based on the relationship between a doorway and a wall. Figure 1 As shown, the indoor-outdoor topological connection method based on the relationship between doorways and walls includes the following steps: Step 110: Obtain the outdoor 3D model and the indoor 3D model, and extract the components of the outdoor 3D model and the indoor 3D model to obtain the indoor and outdoor component set.

[0023] This application extracts a standardized set of components for subsequent geometric analysis by integrating independently constructed outdoor and indoor 3D models and performing coordinate benchmark unification and component-level semantic parsing. The outdoor 3D model is an oblique photogrammetry image in OBJ / OSGB format, or an airborne LiDAR reconstructed mesh, or a point cloud in LAS 1.4 format, with a geodetic coordinate system. The indoor 3D model is a mobile laser scan / panoramic reconstructed point cloud and Scan-to-BIM component model in IFC 4.0 or E57 format, with a local scan coordinate system.

[0024] Step 120: Identify door openings in the set of indoor and outdoor components to obtain a set of candidate door openings.

[0025] This application identifies doorways in a set of indoor and outdoor components by detecting door-shaped openings on both the exterior and interior walls, thus obtaining a candidate doorway set.

[0026] Step 130: For each candidate doorway in the candidate doorway set, determine the attributes of the candidate doorway and filter to obtain the entrance doorway with the attribute of "entrance opening".

[0027] The candidate doorway attributes include entrance opening and open space. This application calculates the normal vector, passage direction, and local geometric description for each candidate doorway, outputting a doorway feature vector carrying the orientation and description, providing a comparable quantitative basis for cross-indoor and outdoor matching.

[0028] Step 140: Based on the distance and location information between the outdoor entrance doorway and the indoor entrance doorway, determine the continuous wall between the outdoor entrance doorway and the indoor entrance doorway and its corresponding confidence level.

[0029] This application examines whether the exterior wall and the interior wall constitute the inner and outer sides of the same solid wall at the opening location along the normal vector direction of each candidate entrance doorway, and calculates the confidence level of the continuity of the continuous wall.

[0030] Step 150: Construct a bipartite graph based on the outdoor entrance doorway and indoor entrance doorway corresponding to the continuous wall.

[0031] In this bipartite graph, the outdoor entrance doorway is the left node and the indoor entrance doorway is the right node. This application constructs a bipartite graph of the correspondence between indoor and outdoor candidate relationships based on continuous wall candidate pairs and the adjacency relationship between indoor room space units and doorways.

[0032] Step 160: Based on the geometric similarity of the door openings, the consistency of their orientation, and the continuity of the walls, solve for the global optimal matching result of the inner and outer door openings on the bipartite graph.

[0033] This application integrates three types of clues—geometric similarity of doorways, consistency of orientation, and continuity of walls—on the basis of candidate bipartite graphs to solve for the globally optimal one-to-one correspondence matching of indoor and outdoor doorways.

[0034] Step 170: Using outdoor and indoor space units as nodes and doorways confirmed by the global optimal matching results as connecting edges, construct a unified indoor and outdoor topology connection diagram.

[0035] This application structures the matching set into a unified indoor and outdoor topology connection graph, and performs a feedback rematching loop on low-confidence candidates to finally obtain the topology data.

[0036] This application provides an indoor-outdoor topological connection method based on the relationship between doorways and walls. The method involves acquiring outdoor and indoor 3D models, extracting components from both models to obtain indoor and outdoor component sets; identifying doorways within these sets to obtain a candidate doorway set; determining the attributes of each candidate doorway and filtering for entrance doorways with the attribute of "entrance opening"; the attributes of candidate doorways include "entrance opening" and "opening into space"; determining the continuous wall between the outdoor and indoor entrance doorways and their corresponding confidence levels based on the distance and location information; constructing a bipartite graph based on the corresponding outdoor and indoor entrance doorways; where the outdoor entrance doorway is the left node and the indoor entrance doorway is the right node; and then, based on the geometric similarity and orientation of the doorways... To ensure consistency and wall continuity, the system solves for the globally optimal matching result of interior and exterior door openings on a bipartite graph. Using exterior and interior space units as nodes and door openings confirmed by the globally optimal matching result as connecting edges, a unified indoor and outdoor topological connection graph is constructed. By constructing an explicit topological connection graph with door openings as connecting edges and interior and exterior space units as nodes, the registration result is upgraded from a simple coordinate transformation to structured topological data that can directly support integrated indoor and outdoor navigation and emergency evacuation simulation. It can automatically complete the determination of indoor and outdoor topological connections without the need for manual pre-calibration of anchor points. Theoretical analysis suggests it can support batch automated processing in urban-level existing building survey scenarios. Furthermore, by combining wall continuity, geometric similarity, and orientation consistency as three criteria for fusion scoring to solve for the globally optimal matching, the matching rate is significantly improved compared to matching methods that rely solely on geometric similarity, thus balancing automation efficiency and critical reliability.

[0037] In one embodiment, step 110 can be implemented as follows: semantic segmentation is performed on the outdoor 3D model to extract outdoor components; wherein, the outdoor components include exterior wall panels, roof panels and ground panels; the component tree is parsed on the indoor 3D model to extract indoor components; wherein, the indoor components include walls, doors, windows and spaces; and the outdoor components and indoor components are stored hierarchically by type as a component attribute table.

[0038] This application employs plane fitting (with an interior point distance threshold of 0.02m and a minimum number of support points of no less than 300) to perform semantic segmentation on the outdoor 3D model, extracting exterior wall panels, roof panels, and ground panels; it parses the IFC component tree of the indoor 3D model, extracting the geometric representations and globally unique identifiers of components such as walls, doors, windows, and spaces; it transforms the indoor local scan coordinate system to the outdoor geodetic coordinate system. If the indoor 3D model lacks absolute positioning information, it uses the building's first-floor elevation and outer contour corner points as control points to perform manual or semi-automatic coarse alignment, with a coarse alignment residual not exceeding 0.5m; it stores the unified coordinate system set of indoor and outdoor components in a standardized component attribute table (including component ID, type, bounding box, normal vector, and floor) hierarchically by type. The attribute table specifically includes the outdoor wall panel set, the indoor wall set, and the indoor room space unit set.

[0039] In one embodiment, step 120 can be implemented as follows: using a boundary hole detection algorithm to identify through holes on the exterior wall panel; performing minimum bounding rectangle fitting on the boundary of the through hole to obtain candidate outdoor door openings; if a door component exists in the interior 3D model, extracting the geometric bounding box of the door component as a candidate indoor door opening; otherwise, performing a hole detection algorithm on the wall to obtain candidate indoor door openings; and obtaining a set of candidate door openings based on the candidate outdoor door openings and the candidate indoor door openings.

[0040] This application uses a boundary hole detection algorithm to identify through holes in the exterior wall panels (excluding those with an area smaller than 1.6m²). 2 or greater than 6m 2 The algorithm detects holes in the walls and performs minimum bounding rectangle fitting on the boundaries of through holes to obtain candidate doorways for the outside (only candidate doorways with a height less than 0.3m from the ground are retained). If a door component exists in the indoor 3D model, the geometric bounding box of the door component is extracted as an indoor candidate doorway; otherwise, a hole detection algorithm is performed on the wall to obtain indoor candidate doorways. Based on the outdoor and indoor candidate doorways, a set of candidate doorways is obtained. Each record in the candidate doorway set includes the center point coordinates, width and height, height from the ground, and the ID of the wall to which it belongs.

[0041] In one embodiment, the specific implementation of step 130 above may be as follows: for each candidate doorway in the candidate doorway set, take the outer normal vector of the wall where the candidate doorway is located as the orientation vector of the candidate doorway, calculate the coordinates of the center point of the doorway, width, height, and height above the ground of the candidate doorway; perform a nearest neighbor radius search with the center point of the doorway as the center of the sphere, and count the wall density distribution in the neighborhood; and distinguish the attributes of the candidate doorway based on the wall density distribution in the neighborhood.

[0042] This application takes the external normal vector of the wall containing the candidate doorway as the orientation vector for each candidate doorway in the candidate doorway set, and calculates the coordinates of the center point, width, height, and ground clearance of the candidate doorway. The direction of the normal vector is determined by the difference in point cloud density on both sides of the wall surface (the side with higher point cloud density is the outside). A nearest neighbor radius search is performed with the center point of the doorway as the center (radius is 1.5m, number of nearest neighbors is 15), and the wall density distribution within the neighborhood is statistically analyzed. Based on the wall density distribution within the neighborhood, the attributes of the candidate doorway are distinguished. For example, the outside of the neighborhood of an open-air doorway has no ground continuity (such as a balcony door), and the outside of the neighborhood of an entrance doorway has passable ground. Furthermore, the candidate set of outdoor doorways of the same building is clustered by orientation (cluster radius is 2m, minimum number of samples is 1), and duplicate screenings of the same entrance doorway caused by scanning segmentation are merged.

[0043] In one embodiment, step 140 can be implemented as follows: calculate the Euclidean distance between the center points of the outdoor entrance doorway and the indoor entrance doorway on the horizontal projection plane; if the Euclidean distance is less than a preset distance threshold, calculate the cosine similarity of the angle between the normal vector of the outdoor wall surface corresponding to the outdoor entrance doorway and the normal vector of the indoor wall corresponding to the indoor entrance doorway; if the cosine similarity is less than a preset similarity threshold, perform ray sampling on the spatial area between the outdoor wall surface and the indoor wall along the direction of the normal vector of the outdoor wall surface, and count the proportion of sampling points falling within a reasonable wall thickness range; if the proportion is not lower than a preset proportion threshold, determine that the spatial area is a continuous wall, and calculate the confidence level of the continuous wall.

[0044] This application calculates the number i The outdoor entrance doorway and the first j The Euclidean distance between the center point of each indoor entrance doorway and its horizontal projection on the plane; if this Euclidean distance is less than a preset distance threshold (default 2m, corresponding to a typical wall thickness of 0.2m to 0.4m plus a doorway positioning error margin), then the first... i The outdoor entrance doorway and the first jOnly candidate pairs consisting of individual indoor entrance doorways are included in the subsequent analysis, thus pruning the search space of candidate correspondences from full pruning to local neighborhood pruning. Then, the cosine similarity of the angle between the normal vector of the outdoor wall surface corresponding to the outdoor entrance doorway and the normal vector of the indoor wall corresponding to the indoor entrance doorway is calculated. If the cosine similarity is less than a preset similarity threshold (-0.85, meaning the angle between the two normal vectors is greater than 148°, approximately opposite, indicating the two walls are on opposite sides of the same wall), they are determined to be facing the same direction. In this case, ray sampling is performed along the direction of the outdoor wall surface normal vector with a step size of 0.05m in the spatial area between the outdoor wall surface and the indoor wall, and the proportion of sampling points falling within a reasonable wall thickness range (0.1m to 0.5m) is counted. If the proportion is not lower than a preset proportion threshold (0.8), the spatial area is determined to be a continuous wall, and the confidence level of the continuous wall is calculated. The formula for calculating the confidence level is: ; in, For the first i The outdoor entrance doorway and the first j The confidence level of the continuity of the walls between indoor entrance doorways. For the first i The outdoor entrance doorway and the first j The European-style distance between the interior entrance doorways For the preset distance threshold, The cosine similarity of the included angle is... The proportion of sampling points falling within a reasonable wall thickness range. These are the weighting coefficients, If the confidence level is less than the set value (0.5), then determine the first... i The outdoor entrance doorway and the first j The walls between the interior entrance doorways are discontinuous.

[0045] In one embodiment, the specific implementation of step 150 above may be: retrieving the room unit adjacent to the indoor entrance doorway corresponding to the continuous wall; if the height of the room unit is within a preset height range, constructing a bipartite graph with the outdoor entrance doorway and the indoor entrance doorway, and establishing a candidate edge for each pair of outdoor entrance doorways and indoor entrance doorways.

[0046] This application retrieves the room units adjacent to the indoor entrance doorway corresponding to the continuous wall. If the height of the room unit is within the preset height range ([-0.15m, 0.15m]), that is, the indoor adjacent room must be basically the same height as the outdoor ground level, excluding misjudgments of non-entry scenarios such as basement light wells and equipment openings, then a bipartite graph is constructed with the outdoor entrance doorway as the left node and the indoor entrance doorway as the right node. A candidate edge is established for each candidate pair, and the initial weight of the candidate edge is the confidence level of the candidate pair. The number of candidate edges for each left node is counted. If a candidate outdoor doorway has no candidate edges (i.e., an isolated node), it is marked as awaiting manual review.

[0047] In one embodiment, step 160 can be implemented as follows: calculate the geometric similarity, orientation consistency score, and comprehensive weight of each candidate edge in the bipartite graph; based on the geometric similarity, orientation consistency score, and comprehensive weight, use the Hungarian algorithm to solve the maximum weight bipartite graph matching problem in the bipartite graph to obtain the matching result; if the comprehensive weight of each edge in the matching result is greater than a preset weight threshold, then the matching result is determined to be the globally optimal matching result.

[0048] This application calculates the geometric similarity, orientation consistency score, and comprehensive weight of each candidate edge in a bipartite graph. Based on the geometric similarity, orientation consistency score, and comprehensive weight, the maximum weight bipartite graph matching is solved using the Hungarian weighted algorithm with the comprehensive weight as the weight, yielding the matching result. If the comprehensive weight of each edge in the matching result is greater than a preset weight threshold, the matching result is determined to be the globally optimal matching result. If the comprehensive weight of an edge in the matching result is less than or equal to a set value (0.6), the matching pair is marked as low confidence and not directly accepted. Steps 140 and 150 are then re-executed (the preset distance threshold can be relaxed to 2.5m, and the preset height range can be relaxed to ±0.2m). If the comprehensive weight is still less than or equal to the set value after re-solving, it is finally marked as awaiting manual review rather than forced matching, to avoid incorrect connections being written into the topology graph. The formula for calculating geometric similarity is: S geo (i,j)=1-(|w_i-w_j| / w_max+|h_i-h_j| / h_max) / 2 ; in, S geo (i, j) For the first i The outdoor entrance doorway and the first j Geometric similarity of interior entrance doorways w_i、 w_j The first i The outdoor entrance doorway and the first j The width of an indoor entrance doorway, h_i、h_j The first i The outdoor entrance doorway and the firstj The height of the indoor entrance doorway w_max =2.4m, h_max =3m.

[0049] The formula for calculating the orientation consistency score is: S dir (i, j) = max(0, -cosθ) ; in, S dir (i, j) For the first i The outdoor entrance doorway and the first j The score for the consistency of orientation of each indoor entrance doorway.

[0050] The formula for calculating the overall weight is: Score(i,j)=α·C wall (i,j)+β·S geo (i,j)+γ·S dir (i, j) ; in, Score(i, j) For the first i The outdoor entrance doorway and the first j The comprehensive weight of the candidate edges corresponding to each indoor entrance doorway. α, β, γ They are weights, α=0.5, β=0.25, γ=0.25 .

[0051] In one embodiment, step 170 can be implemented as follows: using outdoor space units and indoor space units as nodes, and doorways confirmed by the global optimal matching result as connecting edges, a unified indoor and outdoor topology connection diagram can be constructed.

[0052] This application uses outdoor space units (sites, roads, plazas) and indoor space units (rooms, corridors, lobbies) as nodes, and door openings confirmed by the global optimal matching results as connecting edges to construct a unified indoor and outdoor topology connection diagram. Node attributes include space type, floor, and functional category, while edge attributes include door opening width and height, opening direction, and connectivity type (pedestrian / barrier-free / fire evacuation).

[0053] In one embodiment, the above-mentioned indoor-outdoor topology connection method based on the relationship between door openings and walls may further include: searching for paths from each indoor room node to an outdoor node in the unified indoor-outdoor topology connection graph; if all indoor rooms have paths to outdoor nodes, then the final topology connection graph is obtained.

[0054] This application performs connectivity verification on the unified indoor and outdoor topology connection graph, and uses breadth-first search to find the path from each indoor room node to the outdoor node; if all indoor rooms have a path to the outdoor node, the final topology connection graph is obtained; if there are isolated indoor rooms (i.e., a set of rooms that cannot reach the outside), they are marked as topology anomalies and included in the report.

[0055] The following is a specific example illustrating the implementation scheme of this application: Outdoor data was collected using a drone equipped with an oblique photogrammetry camera to capture images of the building facade and surrounding area. The flight path consisted of one circling route around the building and one orthographic route from the top, at an altitude of 40m, a ground resolution of approximately 1.2cm, a directional overlap of 85%, and a lateral overlap of 70%, resulting in 2360 valid images. Simultaneously, a terrestrial 3D laser scanner was used to supplement the scan of the building's ground floor facade and entrance area, with 12 stations and a scan time of approximately 3 minutes per station. The point cloud density was maintained at a distance of 10m with a point spacing of at least 6mm. The outdoor oblique photogrammetry data underwent aerial triangulation and dense reconstruction, outputting an OSGB format real-scene 3D mesh with approximately 28 million faces. This mesh was converted to the engineering coordinate system using seven parameters, with a planar conversion error of ±0.03m. Indoor data was collected using a mobile laser scanning backpack at a scanning speed of approximately 0.8m / s, continuously scanning all floors of the building. A single-floor scan took approximately 25 minutes, outputting an E57 format point cloud with an overall point cloud density of at least 800 points / m². 2 The point cloud density in key areas such as corridors and lobbies shall not be less than 1500 points / m². 2 The software automatically generated semantic modeling of indoor point clouds for walls, floors, ceilings, doors and windows, and output an IFC4.0 format model with a total of approximately 6200 components.

[0056] The OSGB format outdoor mesh and the IFC4.0 format indoor model were imported into the 3D processing platform. RANSAC plane fitting was performed on the outdoor mesh to extract exterior wall patches, with an interior point distance threshold of 0.02m and a minimum supported point count of 300. A total of 47 valid exterior wall patches and 9 roof patches were extracted. The component tree of the IFC model was analyzed, extracting 312 IfcWall components, 68 IfcDoor components, and 134 IfcSpace components. Since the indoor scanning starting point for this project uses the corner of the building's first-floor lobby elevator hall as a local origin and lacks absolute geographic coordinates, the southeast and northwest corner columns of the first-floor lobby were used as two control points for coarse alignment using similarity transformation. After coarse alignment, the residual was 0.31m, meeting the requirement of being within 0.5m. After coordinate unification, the component attribute table was stored in GeoPackage format, with a file size of approximately 340MB.

[0057] Hole detection was performed on 47 outdoor wall panels, and 21 candidate holes were detected, with an area of ​​1.6m².2 Up to 6m 2 After screening based on height from the ground (less than 0.3m), 5 outdoor candidate doorways were retained (including the main entrance, side door, underground garage entrance / exit, and 2 fire escape exits that were mistakenly detected and will be removed later); for the indoor side, since the IFC model already contains IfcDoor components, 68 door component bounding boxes were directly extracted as indoor candidates, and after screening based on height from the ground, 12 candidates adjacent to the exterior wall were retained (including internal room doors that were mistakenly included and will be removed in subsequent steps).

[0058] Orientation normals and geometric descriptions were calculated for 5 outdoor candidates and 12 indoor candidates. A nearest-neighbor radius search was performed with the center of the doorway as the sphere's center and a radius of 1.5m to distinguish between entrance openings and open-air openings. Two fire escape exits were incorrectly detected (their outer neighborhood consisted of fire lane steps rather than a continuous passable ground area; after verification, they were confirmed as emergency exits at elevation differences in the outdoor site rather than regular entrances, and were moved to pending verification rather than being directly accepted). The remaining outdoor candidates were clustered by orientation, and no duplicate candidates were found. Finally, a set of 3 outdoor entrance doorway features was obtained (main entrance, side door, and underground parking garage entrance / exit), and a set of 12 indoor entrance doorway features were obtained for further screening.

[0059] For each of the 3 candidate outdoor entrance doorway features and the 12 candidate indoor entrance doorway features, a total of 36 pairs are formed. First, pruning is performed based on a horizontal projection distance threshold of 2m. The remaining 9 candidate pairs are then used for wall continuity assessment. For each candidate pair, the cosine similarity of the angle between the normal vectors and the percentage of wall thickness sampled along the normal vector direction are calculated. Taking the main entrance and the lobby main entrance as an example: =0.28 m, cosθ=-0.97 (angle approximately 166°), 48 points were sampled along the normal vector direction, of which 39 points fell within the wall thickness range of 0.1m to 0.5m. =0.81; Substitute into the formula to calculate: C wall =0.876, far exceeding the threshold of 0.5, thus classifying it as a continuous wall. After calculating all 9 candidate pairs, 5 candidate pairs... C wall Pairs with a distance greater than 0.5 were retained, while the remaining 4 pairs (most of which were indoor room doors and outdoor candidates that were close in distance but whose normal vector angles did not meet the reverse condition) were eliminated.

[0060] For the five remaining candidate pairs, the room units adjacent to the indoor doorway and their height difference with the ground floor were retrieved. The indoor candidate room corresponding to the underground parking garage entrance was the underground parking garage vestibule on the first basement floor, with a height difference of -2.8m, which exceeded the range of [-0.15m, 0.15m]. This candidate was removed and moved to the next review stage (this opening is actually a staircase downhill entrance, not a direct entrance on the same floor, and manual confirmation is needed to determine whether a cross-floor connection relationship has been established rather than a same-floor topological edge). Finally, a candidate bipartite graph was constructed, with two nodes on the left (outdoor candidates corresponding to the main entrance and side door) and one corresponding candidate on the right, forming two candidate edges with edge weights of 0.876 and 0.712, respectively.

[0061] Calculate the geometric similarity between the two candidate edges. S geo Consistency with orientation S dir Candidate pair for main entrance: Outdoor doorway width 1.82m, height 2.45m; corresponding indoor doorway width 1.8m, height 2.4m. S geo =0.9875; S dir =0.97, Score =0.928. The side door candidate pairs were calculated similarly. Score e=0.803. Solve for the maximum weight matching on this 2×2 sub-bipartite graph (which also contains several isolated nodes that did not pass the preorder criterion), and accept both candidate edges. Score All values ​​exceeded 0.6, and no competitive conflicts were observed. The underground parking garage entrances and exits were moved to the pending review queue and did not participate in this round of automatic matching.

[0062] Using outdoor site nodes, road nodes, and indoor lobby, corridor, and room nodes as graph nodes, and two confirmed matches (main entrance - lobby, side door - stairwell vestibule) as connecting edges, a unified indoor and outdoor topology connection graph is constructed. The total number of nodes is 146 (12 outdoor site / road nodes and 134 indoor room and corridor nodes), with two automatically confirmed connecting edges and one record awaiting verification (underground parking garage entrance / exit). The graph data structure uses an adjacency list for storage. Each node records a unique ID, space type enumeration value, floor, and 3D bounding box. Each edge records the start and end node IDs, doorway width and height, ground clearance, opening direction (inward / outward, read from the IfcDoor's OperationType field), and connectivity type label (pedestrian / accessible / fire evacuation, where edges with a doorway width of not less than 1.2m and no height difference are automatically labeled as accessible). A breadth-first search connectivity check was performed on the graph. Starting from any outdoor node, it was found that, except for the outdoor entrance to the underground parking garage (to be checked), all 133 indoor nodes had at least one path leading to an outdoor node, and no abnormal isolated subgraphs were found (the underground parking garage entrances and exits will be manually confirmed and added as cross-level edges). The entire automated process (excluding manual verification) took approximately 42 seconds on the aforementioned workstation, of which approximately 18 seconds were spent on continuous wall ray sampling, approximately 9 seconds on hole detection, and less than 1 second on algorithm solution (due to the small size of the candidate bipartite graph). During the manual verification phase, the engineer confirmed on the checklist interface that the underground parking garage entrances and exits should be established as cross-level connecting edges between the underground level and the outdoor parking garage ramp, rather than same-level edges. After confirmation, the system automatically added the edge and re-performed the connectivity check, and all 146 nodes could reach the outside, forming a complete topology graph. The final topology diagram is exported in both extended format (with additional doorway width and height, opening direction, and connectivity type attributes) and GeoJSON format, allowing the building's emergency evacuation simulation system to read and use it directly, thus avoiding the extra work of manually drawing indoor and outdoor connectivity diagrams again.

[0063] To verify the practical effect of the method of this invention, the project team arranged for two engineers with 3D modeling experience to manually compare the entrance correspondence between the outdoor oblique photogrammetry model and the indoor BIM model in professional modeling software using traditional manual comparison methods, and manually draw the connection relationships as a comparison baseline. Manually completing the verification, coordinate checking, and connection relationship drawing of the three external openings of the same building took approximately 3.2 hours (including mutual verification), with about 40% of the time spent on back-and-forth comparison of the coordinates of the indoor and outdoor models and the doorway positions. Using the method of this application, except for the control point selection in the coarse coordinate alignment step which requires manual intervention (approximately 6 minutes), all other steps are completed automatically. Including the manual confirmation in the verification step (approximately 2 minutes), the total process time is approximately 50 minutes, with calculation time accounting for less than 2%, and the remainder being the waiting time for data format conversion and manual confirmation. Based on the above actual measurement comparison, the manual time in this embodiment is reduced by approximately 74%. In terms of accuracy, the two automatically confirmed connection edges output by the automated process were verified by engineers to be correct (100% accuracy). One record to be verified (underground parking garage entrance / exit) was also correctly identified as a special scenario that required cross-layer processing rather than being incorrectly forcibly matched as a same-layer connection. This indicates that the confidence threshold and feedback mechanism did not produce false positive connections in this embodiment scenario.

[0064] Figure 2 This is a schematic diagram of an indoor-outdoor topological connection system based on the relationship between a doorway and a wall, provided in an exemplary embodiment of this application. Figure 2As shown, the indoor-outdoor topology connection system 20 based on the relationship between doorways and walls includes: a component set extraction module 21, used to acquire outdoor 3D models and indoor 3D models, and extract components from the outdoor and indoor 3D models to obtain indoor and outdoor component sets; a candidate doorway identification module 22, used to identify doorways in the indoor and outdoor component sets to obtain a candidate doorway set; an entrance doorway determination module 23, used to determine the attributes of each candidate doorway in the candidate doorway set and filter to obtain entrance doorways with the attribute of entrance opening; wherein, the attributes of the candidate doorways include entrance opening and open space opening; and a continuous wall determination module 24, used to determine the continuous wall based on the relationship between outdoor and indoor entrance doorways. Distance and location information are used to determine the continuous wall between the outdoor and indoor entrance doors and their corresponding confidence levels; the bipartite graph construction module 25 is used to construct a bipartite graph based on the outdoor and indoor entrance doors corresponding to the continuous wall; wherein the bipartite graph has the outdoor entrance door as the left node and the indoor entrance door as the right node; the optimal matching solution module 26 is used to solve the global optimal matching result of the inner and outer door openings on the bipartite graph based on the geometric similarity of the door openings, the consistency of orientation, and the continuity of the wall; the topology graph construction module 27 is used to construct a unified indoor and outdoor topology connection graph with the outdoor space unit and the indoor space unit as nodes and the door openings confirmed by the global optimal matching result as connecting edges.

[0065] This application provides an indoor-outdoor topological connection system based on the relationship between doorways and walls. A component set extraction module 21 acquires outdoor and indoor 3D models and extracts components from them to obtain indoor and outdoor component sets. A candidate doorway identification module 22 identifies doorways in the indoor and outdoor component sets to obtain a candidate doorway set. An entrance doorway determination module 23 determines the attributes of each candidate doorway in the candidate doorway set and filters them to obtain entrance doorways with the attribute of "entrance opening." The attributes of the candidate doorways include "entrance opening" and "opening into space." A continuous wall determination module 24 determines the continuous wall located between the outdoor and indoor entrance doorways and its corresponding confidence level based on the distance and location information between the outdoor and indoor entrance doorways. A bipartite graph construction module 25 constructs a bipartite graph based on the outdoor and indoor entrance doorways corresponding to the continuous wall. The bipartite graph uses the outdoor entrance doorway as the left node and the indoor entrance doorway as the right node. The optimal matching solution module 26 solves for the global optimal matching result of the inner and outer door openings on the bipartite graph based on the geometric similarity of the door openings, the consistency of orientation, and the continuity of the walls. The topology graph construction module 27 constructs a unified indoor and outdoor topology connection graph with the outdoor space unit and the indoor space unit as nodes and the door openings confirmed by the global optimal matching result as connecting edges. By constructing an explicit topology connection graph with the door openings as connecting edges and the indoor and outdoor space units as nodes, the registration result is upgraded from a simple coordinate transformation to structured topology data that can directly support integrated indoor and outdoor navigation and emergency evacuation simulation. It can automatically complete the determination of indoor and outdoor topology connections without the need for manual pre-calibration of anchor points. Theoretical analysis suggests that it can support batch automated processing in the scenario of city-level existing building survey. Furthermore, by combining the three clues of wall continuity, geometric similarity, and orientation consistency to solve for the global optimal matching, the matching rate is greatly improved compared with the matching method that only relies on a single geometric similarity, thus balancing automation efficiency and evaluation reliability.

[0066] In one embodiment, the component set extraction module 21 can be further configured to: perform semantic segmentation on the outdoor 3D model to extract outdoor components; wherein, the outdoor components include exterior wall panels, roof panels and ground panels; parse the component tree of the indoor 3D model to extract indoor components; wherein, the indoor components include walls, doors, windows and spaces; and store the outdoor components and indoor components in a component attribute table according to their types.

[0067] In one embodiment, the candidate door opening identification module 22 can be further configured to: identify through holes on the exterior wall panel using a boundary hole detection algorithm; perform minimum bounding rectangle fitting on the boundary of the through hole to obtain outdoor candidate door openings; if there is a door component in the indoor 3D model, extract the geometric bounding box of the door component as an indoor candidate door opening; otherwise, perform a hole detection algorithm on the wall to obtain indoor candidate door openings; and obtain a candidate door opening set based on the outdoor candidate door openings and the indoor candidate door openings.

[0068] In one embodiment, the aforementioned entrance doorway determination module 23 may be further configured to: for each candidate doorway in the candidate doorway set, take the outer normal vector of the wall where the candidate doorway is located as the orientation vector of the candidate doorway, calculate the coordinates of the center point of the doorway, width, height, and height above the ground of the candidate doorway; perform a nearest neighbor radius search with the center point of the doorway as the center of the sphere, and statistically analyze the wall density distribution in the neighborhood; and distinguish the attributes of the candidate doorway based on the wall density distribution in the neighborhood.

[0069] In one embodiment, the continuous wall determination module 24 can be further configured to: calculate the Euclidean distance between the center points of the outdoor entrance doorway and the indoor entrance doorway on the horizontal projection plane; if the Euclidean distance is less than a preset distance threshold, calculate the cosine similarity of the angle between the normal vector of the outdoor wall surface corresponding to the outdoor entrance doorway and the normal vector of the indoor wall corresponding to the indoor entrance doorway; if the cosine similarity is less than a preset similarity threshold, perform ray sampling on the spatial area between the outdoor wall surface and the indoor wall along the direction of the normal vector of the outdoor wall surface, and count the proportion of sampling points falling within a reasonable wall thickness range; if the proportion is not lower than a preset proportion threshold, determine the spatial area as a continuous wall and calculate the confidence level of the continuous wall.

[0070] In one embodiment, the bipartite graph construction module 25 can be further configured to: retrieve the room units adjacent to the indoor entrance doorway corresponding to the continuous wall; if the height of the room unit is within a preset height range, construct a bipartite graph with the outdoor entrance doorway and the indoor entrance doorway, and establish a candidate edge for each pair of outdoor entrance doorways and indoor entrance doorways.

[0071] In one embodiment, the optimal matching solution module 26 can be further configured to: calculate the geometric similarity, orientation consistency score and comprehensive weight of each candidate edge in the bipartite graph; based on the geometric similarity, orientation consistency score and comprehensive weight, use the Hungarian algorithm to solve the maximum weight bipartite graph matching in the bipartite graph to obtain the matching result; if the comprehensive weight of each edge in the matching result is greater than the preset weight threshold, then the matching result is determined to be the global optimal matching result.

[0072] In one embodiment, the topology graph construction module 27 can be further configured to: construct a unified indoor and outdoor topology connection graph with outdoor space units and indoor space units as nodes and doorways confirmed by the global optimal matching result as connecting edges.

[0073] In one embodiment, the above-mentioned indoor-outdoor topology connection system 20 based on the relationship between door openings and walls can be further configured to: search for paths from each indoor room node to an outdoor node in the unified indoor-outdoor topology connection graph; if all indoor rooms have paths to outdoor nodes, then the final topology connection graph is obtained.

[0074] Below, for reference Figure 3 To describe an electronic device according to embodiments of this application. For example... Figure 3 As shown, the electronic device 10 includes one or more processors 11 and memory 12.

[0075] The processor 11 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

[0076] The memory 12 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may execute the program instructions to implement the methods of the various embodiments of this application described above and / or other desired functions. Various contents such as input signals, signal components, and noise components may also be stored in the computer-readable storage medium.

[0077] In one example, the electronic device 10 may also include an input device 13 and an output device 14, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0078] When the electronic device is a standalone device, the input device 13 can be a communication network connector for receiving the collected input signals from the first device and the second device.

[0079] In addition, the input device 13 may also include, for example, a keyboard, a mouse, etc.

[0080] The output device 14 can output various information to the outside, including determined distance information, direction information, etc. The output device 14 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0081] Of course, for the sake of simplicity, Figure 3 Only some of the components of the electronic device 10 relevant to this application are shown in this illustration; components such as buses, input / output interfaces, etc., are omitted. In addition, the electronic device 10 may include any other suitable components depending on the specific application.

[0082] In addition to the methods and apparatus described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of this application described in the "Exemplary Methods" section above.

[0083] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0084] Furthermore, embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of this application described in the "Exemplary Methods" section above.

[0085] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0086] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0087] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0088] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0089] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0090] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for connecting indoor and outdoor topologies based on the relationship between doorways and walls, characterized in that, include: Obtain an outdoor 3D model and an indoor 3D model, and extract the components of the outdoor 3D model and the indoor 3D model to obtain an indoor and outdoor component set; Identify door openings in the set of indoor and outdoor components to obtain a candidate door opening set; For each candidate doorway in the candidate doorway set, the attributes of the candidate doorway are determined and the entrance doorway with the attribute of "entrance opening" is obtained; wherein, the attributes of the candidate doorway include "entrance opening" and "opening to the air"; Based on the distance and location information between the outdoor entrance doorway and the indoor entrance doorway, the continuous wall between the outdoor entrance doorway and the indoor entrance doorway and the corresponding confidence level are determined; A bipartite graph is constructed based on the outdoor entrance doorway and the indoor entrance doorway corresponding to the continuous wall; wherein the bipartite graph has the outdoor entrance doorway as the left node and the indoor entrance doorway as the right node; Based on the geometric similarity of door openings, consistency of orientation, and continuity of walls, the global optimal matching result of inner and outer door openings is solved on the bipartite graph; Using outdoor and indoor space units as nodes and doorways confirmed by the global optimal matching results as connecting edges, a unified indoor and outdoor topology connection diagram is constructed. Determining the attributes of each candidate doorway in the candidate doorway set includes: For each candidate doorway in the candidate doorway set, the outer normal vector of the wall where the candidate doorway is located is taken as the orientation vector of the candidate doorway, and the coordinates of the center point, width, height and height above the ground of the candidate doorway are calculated. Using the center point of the doorway as the center of the sphere, a nearest neighbor radius search is performed, and the wall density distribution within the neighborhood is statistically analyzed; The attributes of the candidate doorway are distinguished based on the wall density distribution within the neighborhood.

2. The indoor-outdoor topological connection method based on the relationship between doorways and walls according to claim 1, characterized in that, The extraction of components from the outdoor 3D model and the indoor 3D model to obtain an indoor and outdoor component set includes: Semantic segmentation is performed on the outdoor 3D model to extract outdoor components; wherein, the outdoor components include exterior wall panels, roof panels, and ground panels; The component tree of the interior 3D model is parsed to extract interior components; wherein, the interior components include walls, doors, windows, and spaces; The outdoor components and the indoor components are stored hierarchically by type as a component attribute table.

3. The indoor-outdoor topological connection method based on the relationship between doorways and walls according to claim 2, characterized in that, The process of identifying doorways in the set of indoor and outdoor components to obtain a candidate doorway set includes: A boundary hole detection algorithm is used to identify through holes on the exterior wall panel. Perform minimum bounding rectangle fitting on the boundary of the through hole to obtain candidate outdoor doorways; If the interior 3D model contains a door component, the geometric bounding box of the door component is extracted as a candidate door opening; otherwise, a hole detection algorithm is performed on the wall to obtain candidate door openings. Based on the outdoor candidate doorway and the indoor candidate doorway, the candidate doorway set is obtained.

4. The indoor-outdoor topological connection method based on the relationship between doorways and walls according to claim 3, characterized in that, The determination of the continuous wall between the outdoor entrance doorway and the indoor entrance doorway and its corresponding confidence level based on the distance and location information of the outdoor entrance doorway and the indoor entrance doorway includes: Calculate the Euclidean distance between the center points of the outdoor entrance doorway and the indoor entrance doorway on the horizontal projection plane; If the Euclidean distance is less than a preset distance threshold, then the cosine similarity of the angle between the normal vector of the outdoor wall surface corresponding to the outdoor entrance doorway and the normal vector of the indoor wall corresponding to the indoor entrance doorway is calculated. If the cosine similarity of the included angle is less than the preset similarity threshold, then ray sampling is performed on the spatial area between the outdoor wall panel and the indoor wall along the direction of the normal vector of the outdoor wall panel, and the proportion of sampling points falling within the reasonable wall thickness range is counted. If the ratio is not lower than a preset ratio threshold, the space area is determined to be a continuous wall, and the confidence level of the continuous wall is calculated.

5. The indoor-outdoor topological connection method based on the relationship between doorways and walls according to claim 4, characterized in that, The construction of a bipartite graph based on the outdoor and indoor entrance doorways corresponding to the continuous wall includes: Retrieve the room unit adjacent to the indoor entrance doorway corresponding to the continuous wall; If the height of the room unit is within a preset height range, a bipartite graph is constructed using the outdoor entrance doorway and the indoor entrance doorway, and a candidate edge is created for each pair of outdoor entrance doorways and indoor entrance doorways.

6. The indoor-outdoor topological connection method based on the relationship between doorways and walls according to claim 5, characterized in that, The global optimal matching result for inner and outer door openings on the bipartite graph, based on the geometric similarity of door openings, consistency of orientation, and continuity of walls, includes: Calculate the geometric similarity, orientation consistency score, and overall weight of each candidate edge in the bipartite graph; Based on the geometric similarity, the orientation consistency score, and the comprehensive weight, the Hungarian algorithm is used to solve the maximum weight bipartite graph matching on the bipartite graph to obtain the matching result. If the combined weight of each edge in the matching result is greater than a preset weight threshold, then the matching result is determined to be the globally optimal matching result.

7. The indoor-outdoor topological connection method based on the relationship between doorways and walls according to claim 6, characterized in that, The construction of a unified indoor-outdoor topology connection diagram, using outdoor and indoor space units as nodes and doorways confirmed by the global optimal matching result as connecting edges, includes: Using the outdoor space unit and the indoor space unit as nodes, and the doorway confirmed by the global optimal matching result as the connecting edge, a unified indoor and outdoor topology connection diagram is constructed.

8. The indoor-outdoor topological connection method based on the relationship between doorways and walls according to claim 1, characterized in that, The indoor-outdoor topological connection method based on the relationship between doorways and walls also includes: Search the unified indoor and outdoor topology connection graph for the path from each indoor room node to the outdoor node; If all indoor rooms have paths leading to outdoor nodes, then the final topology graph is obtained.

9. An indoor-outdoor topological connection system based on the relationship between doorways and walls, characterized in that, include: The component set extraction module is used to acquire the outdoor 3D model and the indoor 3D model, and extract the components of the outdoor 3D model and the indoor 3D model to obtain the indoor and outdoor component sets; The candidate doorway recognition module is used to identify doorways in the set of indoor and outdoor components to obtain a candidate doorway set; An entrance doorway determination module is used to determine the attributes of each candidate doorway in the candidate doorway set and filter to obtain entrance doorways with the attribute of "entrance opening"; wherein, the attributes of the candidate doorway include "entrance opening" and "opening to the air"; The continuous wall determination module is used to determine the continuous wall located between the outdoor entrance doorway and the indoor entrance doorway and the corresponding confidence level based on the distance and location information between the outdoor entrance doorway and the indoor entrance doorway. The bipartite graph construction module is used to construct a bipartite graph based on the outdoor entrance doorway and the indoor entrance doorway corresponding to the continuous wall; wherein the bipartite graph has the outdoor entrance doorway as the left node and the indoor entrance doorway as the right node; The optimal matching solution module is used to solve the global optimal matching result of the inner and outer door openings on the bipartite graph based on the geometric similarity of the door openings, the consistency of the orientation, and the continuity of the walls; The topology graph construction module is used to construct a unified indoor and outdoor topology connection graph with outdoor space units and indoor space units as nodes and doorways confirmed by the global optimal matching result as connecting edges. The entrance doorway determination module is further configured as follows: For each candidate doorway in the candidate doorway set, the outer normal vector of the wall where the candidate doorway is located is taken as the orientation vector of the candidate doorway, and the coordinates of the center point, width, height and height above the ground of the candidate doorway are calculated. Using the center point of the doorway as the center of the sphere, a nearest neighbor radius search is performed, and the wall density distribution within the neighborhood is statistically analyzed; The attributes of the candidate doorway are distinguished based on the wall density distribution within the neighborhood.

Images