A three-dimensional building model management system and method based on modular design

Abstract

The application relates to the field of building models, in particular to a three-dimensional building model management system and method based on modular design, which comprises a point cloud forming module, a component management module, a assembling and integrating module, an internal expansion module and a model checking module; the point cloud forming module is used for collecting entity point clouds and generating building component models; the component management module is used for establishing topological connection relationships among components; the assembling and integrating module is used for obtaining building external models; the internal expansion module is used for establishing in-layer topological road networks; and the model checking module is used for merging and outputting three-dimensional building models; the application can improve the component modeling efficiency, improve the model details, reduce the building cost, solve the problems of low three-dimensional modeling automation, unreal texture, poor model regularity and the like, make the model have richer semantic information analysis capability, improve the model reality, and have good expansibility and practicality.

Description

Technical Field

[0001] This invention relates to the field of architectural models, specifically to a three-dimensional architectural model management system and method based on modular design. Background Technology

[0002] Building Information Modeling (BIM) is a technology that uses three-dimensional digital models of buildings for engineering data management. By creating virtual building models, it enables sustainable management of buildings throughout their entire lifecycle, greatly improving the efficiency of building information utilization. Three-dimensional building models constructed using BIM technology typically feature complex spatial structures, massive data volumes, and rich textural details. For large and complex buildings, the building model often cannot be completed in one go and requires a modular assembly approach to ensure the model's construction accuracy.

[0003] As the number of BIM modular components continues to increase and architectural designs become increasingly personalized, existing component elements are difficult to store and reuse in large quantities, making it challenging to build BIM prefabricated models. Furthermore, because point clouds themselves lack semantic richness, using point cloud scanning to import entity components into the database can easily result in distortions such as irregular edges, low geometric accuracy, and uneven surfaces, affecting the geometric integrity of the model.

[0004] Furthermore, due to the large number of cross-floor components with irregular geometric shapes, the data quality distribution inside the building is uneven, often resulting in problems such as occlusion and missing local information. This leads to the loss of local structure and geometric accuracy of the 3D building model constructed from prefabricated components. Modular building semantic extraction methods are not very practical, internal model construction is difficult, and the model rendering process is costly. Summary of the Invention

[0005] The purpose of this invention is to provide a three-dimensional building model management system and method based on modular design to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a three-dimensional building model management system based on modular design, comprising: a point cloud forming module, a component management module, an assembly and merging module, an internal expansion module, and a model verification module; The point cloud forming module is used to collect point cloud data of building module entities. Based on the random forest algorithm, it classifies the point cloud according to the geometric and spectral features, calculates the features in the neighborhood of each point, separates the auxiliary point cloud from the building point cloud, extracts the point cloud into individual entities through the density clustering extraction module, extracts the view frustum plane from the point cloud, calculates the spatial corner points of the plane intersection line, uses the intersection judgment algorithm to filter the closed space, establishes the bounding box space, and divides the point cloud into independent physical components to obtain the building component model. The component management module is used to process the vertex features of the model based on vertex constraints and angle error weighting, perform vertex constraint adjustment through angle error weighting, calculate the outer bounding box center of the building component model with equal latitude and longitude difference grid, determine the center coordinates of the building component model, import all building component models into the modular component library with the center coordinates as index, divide the logical components in the library according to building function or volume, establish the topological connection relationship between components, and maintain the associated boundary when the component boundary changes. The assembly and merging module is used to build a three-dimensional building model using building components. It aligns the component point cloud in the weighted principal direction of the building assembly constraints, determines the module scaling height based on the normal direction and centroid of the building surface, merges building edge features using line fitting technology, reconstructs the model surface using a triangular mesh reconstruction algorithm, performs texture mapping based on the building model texture library, corrects erroneous planar topology connections, deletes redundant corner points, and obtains the external model. The internal expansion module is used to divide the internal area of ​​the building according to the flatness by using the region growing algorithm, and to stitch and register the texture images by remapping the interlayer features of the internal wall surface and according to the scale-invariant feature transformation of the texture, remove texture noise, and apply footprint simulation under the constraints of building rules by utilizing the distribution of the intralayer area to establish a topological road network. The model verification module is used to project cross-floor components onto a plane, extract two-dimensional boundaries from the footprint point cloud using a boundary extraction algorithm, identify inter-floor connection points as inter-floor traffic entrances, plan inter-floor road network topology paths, link inter-floor road networks and topology road networks to obtain a hybrid road network model, merge the external model and the hybrid road network model, output a three-dimensional building information model, and verify the building structure based on vertex smoothness and texture features.

[0007] Furthermore, the point cloud forming module includes: a data acquisition unit, a point cloud classification unit, and a single-unit extraction unit; The data acquisition unit is used to acquire laser point clouds and densely matched point clouds of prefabricated buildings using a TLS scanner or oblique photogrammetry system, and then register and fuse them to obtain a point cloud dataset. The point cloud classification unit is used to uniformly downsample the point cloud dataset using a voxelized grid method to remove outliers and scanning noise, and to identify point cloud features, including: linearity, planarity, scattering, curvature, elevation, and point density. The single-unit extraction unit is used to train a random forest classifier using labeled samples, separate the auxiliary point cloud introduced by non-building features, extract building facades using density clustering algorithm, merge homogeneous regions, and output building component models.

[0008] Furthermore, the component management module includes: a vertex optimization unit and an indexing unit; The vertex optimization unit is used to identify model vertices from corner points. The vertices are located on the plane of the model's exterior facade or on the intersection of planes, and the adjacent edges connected to the vertices are perpendicular or parallel to each other. The inbound index unit is used to establish a spatial index, determine the grid number by the center coordinates of the component, and import a modular component library containing geometric information, attribute information, spatial information, and connection relationship information.

[0009] Furthermore, the assembly and merging module includes: an intelligent assembly unit, a geometric reconstruction unit, and a texture mapping unit; The intelligent assembly unit is used to extract building component models from the component library, calculate the dominant axis of the building as a whole on the horizontal plane according to the set building assembly requirements, and rotate all component models around the center of the dominant axis to align the horizontal axis of the building coordinate system with the weighted principal direction. The geometric reconstruction unit is used to establish the connection relationship between components. Taking the normal of the building floor plan or ground as the reference plane, it calculates the vertical distance from the centroid of the point cloud of the aligned component to the reference plane, compares the ratio of the vertical distance to the input standard floor height, and calculates the scaling factor of each component model. The texture mapping unit is used to extract the building outline texture from the aligned component model. By least squares straight line fitting, textures with the same straight line features are fitted into continuous contour lines. Based on the model unfolding algorithm, texture coordinates are calculated for each contour surface, and the correspondence between mesh vertices and texture image pixels is established.

[0010] Furthermore, the internal expansion module includes: a region segmentation unit and a road network planning unit; The region segmentation unit is used to extract the vertices of the inner surface triangular mesh from the external model, select points located on the continuous plane as initial seeds, and classify points whose normal vector angle and distance to the plane are both less than the threshold into the same region, thereby dividing the interior space of the building into different semantic regions. The road network planning unit is used to identify road connections, including doors, internal staircases and openings, by using straight line fitting and intersection calculation. The center point of the connection is used as a node, and the center line connecting the nodes is used as an edge, thereby establishing a topology map inside each floor and generating a polygonal road network in the plane of each floor.

[0011] Furthermore, the model verification module includes: an inter-layer planning unit and a LOD output unit; The inter-floor planning unit is used to project cross-floor components as inter-floor connection points. The cross-floor components include stairs, elevator shafts, and ramps. The inter-floor connection points are added to the topology diagrams of the upper and lower floors respectively to create a high-rise hybrid road network model. The LOD output unit is used to unify the model reference system, associate the nodes and edges in the road network with the geometric elements in the external model, output a three-dimensional building information model, and perform local curvature analysis and texture feature verification on the model.

[0012] A method for managing 3D building models based on modular design includes the following steps: Step S1. Collect point cloud data of building module entities. Based on the geometric and spectral characteristics of the point cloud, separate the auxiliary point cloud from the building point cloud. Extract the view frustum plane from the building point cloud, establish the bounding box space, and divide the point cloud into independent physical components to obtain the building component model. Step S2. Calculate the center coordinates of each building component model. Using the center coordinates as an index, import all building component models into the component library, classify logical components according to building function or volume, and establish the topological connection relationship between components. Step S3. Use building components to build a three-dimensional building model, align the component point cloud in the weighted principal direction of the building assembly, scale each component according to the normal direction and center of gravity of the building surface, merge the building edge features, perform texture mapping based on the exterior walls and common contours, reconstruct the model surface, and obtain the building exterior model. Step S4. Segment the internal areas of the building, remap the inter-layer features of the internal walls, establish a road network planning model using the distribution data of the intra-layer areas, apply footprint simulation under the constraints of building rules, and establish an intra-layer topological road network; Step S5. Project the cross-floor components onto the plane, use the inter-floor connection points as cross-floor traffic entrances, establish an inter-floor road network, link the inter-floor road network and the topological road network to obtain a hybrid road network model, merge the building exterior model and the hybrid road network model, output a three-dimensional building information model, and verify the building structure based on vertex smoothness and texture features.

[0013] Furthermore, step S1 includes: Step S11. Use a TLS scanner or oblique photogrammetry system to collect laser point clouds and densely matched point clouds of prefabricated buildings, register and fuse them to obtain a point cloud dataset, uniformly downsample the point cloud dataset using the voxelized mesh method to remove outliers and scanning noise, and identify point cloud features, including: linearity, planarity, scattering, curvature, elevation and point density. Step S12. Use labeled samples to train a random forest classifier, separate the auxiliary point cloud introduced by non-building features, use density clustering algorithm to extract building facades, merge homogeneous areas, calculate the spatial corner points of the plane intersection line, use the intersection judgment algorithm to filter closed spaces, and output the building component model.

[0014] Furthermore, step S2 includes: Step S21. Based on vertex constraints and angle error weighted processing of model vertex features, identify model vertices from corner points. The vertices are located on the plane of the model's exterior facade or on the plane intersection line, and the adjacent edges connected to the vertices are perpendicular or parallel. Vertex constraint adjustment is performed through angle error weighting. The center of the outer bounding box of the building component model is calculated using an equal latitude and longitude difference grid to determine the center coordinates of the building component model. Step S22. Establish a spatial index, determine the grid number based on the center coordinates of the component, import a modular component library containing geometric information, attribute information, spatial information and connection relationship information, establish topological connection relationships between components, and maintain associated boundaries when component boundaries change.

[0015] Furthermore, step S3 includes: Step S31. Extract building component models from the component library, calculate the dominant axis of the building as a whole on the horizontal plane according to the set building assembly requirements, and rotate all component models around the center of the dominant axis to align the horizontal axis of the building coordinate system with the weighted principal direction. Step S32. Establish the connection relationship between components. Using the normal of the building floor plan or ground as the reference plane, calculate the vertical distance from the centroid of the point cloud of the aligned components to the reference plane. Compare the ratio of the vertical distance to the input standard floor height and calculate the scaling factor of each component model. Step S33. Extract the building outline texture from the aligned component model. By least squares straight line fitting, fit textures with the same straight line features into a continuous outline. Based on the model unfolding algorithm, calculate the texture coordinates for each outline surface and establish the correspondence between mesh vertices and texture image pixels.

[0016] Furthermore, step S4 includes: Step S41. Extract the vertices of the inner surface triangular mesh from the external model, select the points located on the continuous plane as the initial seed, and classify the points whose normal vector angle and distance to the plane are both less than the threshold into the same region, thus dividing the interior space of the building into different semantic regions. Step S42. Using line fitting and intersection calculation, identify road connections, including doors, internal staircases and openings. Use the center point of the connection as a node and the center line connecting the nodes as an edge to establish the topology map inside each floor and generate a polygonal road network in the plane of each floor.

[0017] Furthermore, step S5 includes: Step S51. Project cross-floor components as inter-floor connection points. The cross-floor components include stairs, elevator shafts and ramps. Use a boundary extraction algorithm to extract two-dimensional boundaries from the footprint point cloud. Add the inter-floor connection points to the topology maps of the upper and lower floors respectively, and plan the inter-floor road network topology path. Step S52. Unify the model reference system, associate the nodes and edges in the road network with the geometric elements in the external model, output the three-dimensional building information model, and perform local curvature analysis and texture feature verification on the model.

[0018] Compared with the prior art, the beneficial effects achieved by the present invention are: 1. This invention collects point cloud data of building module entities, establishes bounding box space, determines the center coordinates of building component models using equal latitude and longitude difference grids, imports the data into a modular component library, and performs segmentation and boundary maintenance. This effectively improves modeling speed, is suitable for scenarios where the component model itself has a large number of triangular faces, improves the efficiency of component modeling, enhances model details, improves the level of parametric modeling, reduces construction costs, and ensures the geometric regularity of the building model.

[0019] 2. This invention can build a three-dimensional building model using building components, align the component point clouds in the main direction, merge building edge features, reconstruct the model surface, correct erroneous planar topology connections, and delete redundant corner points to obtain the external building model. It realizes the automated reconstruction from components to a three-dimensional building model, improves the model's standardization, makes the model more lightweight, and solves problems such as low automation in three-dimensional modeling, unrealistic textures, and poor model regularity.

[0020] 3. This invention can segment the internal area of ​​a building according to flatness, remove texture noise, plan the inter-layer road network topology path, connect the inter-layer road network and the topology road network, merge the external model and the hybrid road network, and output the building model. This enables the model to have richer semantic information and computational analysis capabilities, meet the needs of refined application scenarios such as building digital twins, improve the model's realism and visual expressiveness, and has good scalability and practicality. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the structure of a three-dimensional building model management system based on modular design according to the present invention; Figure 2 This is a schematic diagram illustrating the steps of a modular design-based three-dimensional building model management method according to the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figures 1 to 2 The present invention provides a technical solution: a three-dimensional building model management system based on modular design, comprising: a point cloud forming module, a component management module, an assembly and merging module, an internal expansion module, and a model verification module; The point cloud forming module is used to collect point cloud data of building module entities. Based on the random forest algorithm, it classifies the point cloud according to the geometric and spectral features, calculates the features in the neighborhood of each point, separates the auxiliary point cloud from the building point cloud, extracts the point cloud into individual entities through the density clustering extraction module, extracts the view frustum plane from the point cloud, calculates the spatial corner points of the plane intersection line, uses the intersection judgment algorithm to filter the closed space, establishes the bounding box space, and divides the point cloud into independent physical components to obtain the building component model. The point cloud forming module includes: a data acquisition unit, a point cloud classification unit, and a single-unit extraction unit; The data acquisition unit is used to acquire laser point clouds and densely matched point clouds of prefabricated buildings using a TLS scanner or oblique photogrammetry system, and then register and fuse them to obtain a point cloud dataset. The point cloud classification unit is used to uniformly downsample the point cloud dataset using a voxelized grid method to remove outliers and scanning noise, and to identify point cloud features, including: linearity, planarity, scattering, curvature, elevation, and point density. The single-unit extraction unit is used to train a random forest classifier using labeled samples, separate the auxiliary point cloud introduced by non-building features, extract building facades using density clustering algorithm, merge homogeneous regions, and output building component models.

[0024] The component management module is used to process the vertex features of the model based on vertex constraints and angle error weighting, perform vertex constraint adjustment through angle error weighting, calculate the outer bounding box center of the building component model with equal latitude and longitude difference grid, determine the center coordinates of the building component model, import all building component models into the modular component library with the center coordinates as index, divide the logical components in the library according to building function or volume, establish the topological connection relationship between components, and maintain the associated boundary when the component boundary changes. The component management module includes: a vertex optimization unit and an indexing unit; The vertex optimization unit is used to identify model vertices from corner points. The vertices are located on the plane of the model's exterior facade or on the intersection of planes, and the adjacent edges connected to the vertices are perpendicular or parallel to each other. The inbound index unit is used to establish a spatial index, determine the grid number by the center coordinates of the component, and import a modular component library containing geometric information, attribute information, spatial information, and connection relationship information.

[0025] The assembly and merging module is used to build a three-dimensional building model using building components. It aligns the component point cloud in the weighted principal direction of the building assembly constraints, determines the module scaling height based on the normal direction and centroid of the building surface, merges building edge features using line fitting technology, reconstructs the model surface using a triangular mesh reconstruction algorithm, performs texture mapping based on the building model texture library, corrects erroneous planar topology connections, deletes redundant corner points, and obtains the external model. The assembly and merging module includes: an intelligent assembly unit, a geometric reconstruction unit, and a texture mapping unit; The intelligent assembly unit is used to extract building component models from the component library, calculate the dominant axis of the building as a whole on the horizontal plane according to the set building assembly requirements, and rotate all component models around the center of the dominant axis to align the horizontal axis of the building coordinate system with the weighted principal direction. The geometric reconstruction unit is used to establish the connection relationship between components. Taking the normal of the building floor plan or ground as the reference plane, it calculates the vertical distance from the centroid of the point cloud of the aligned component to the reference plane, compares the ratio of the vertical distance to the input standard floor height, and calculates the scaling factor of each component model. The texture mapping unit is used to extract the building outline texture from the aligned component model. By least squares straight line fitting, textures with the same straight line features are fitted into continuous contour lines. Based on the model unfolding algorithm, texture coordinates are calculated for each contour surface, and the correspondence between mesh vertices and texture image pixels is established.

[0026] The internal expansion module is used to divide the internal area of ​​the building according to the flatness by using the region growing algorithm, and to stitch and register the texture images by remapping the interlayer features of the internal wall surface and according to the scale-invariant feature transformation of the texture, remove texture noise, and apply footprint simulation under the constraints of building rules by utilizing the distribution of the intralayer area to establish a topological road network. The internal expansion module includes: a region segmentation unit and a road network planning unit; The region segmentation unit is used to extract the vertices of the inner surface triangular mesh from the external model, select points located on the continuous plane as initial seeds, and classify points whose normal vector angle and distance to the plane are both less than the threshold into the same region, thereby dividing the interior space of the building into different semantic regions. The road network planning unit is used to identify road connections, including doors, internal staircases and openings, by using straight line fitting and intersection calculation. The center point of the connection is used as a node, and the center line connecting the nodes is used as an edge, thereby establishing a topology map inside each floor and generating a polygonal road network in the plane of each floor.

[0027] The model verification module is used to project cross-floor components onto a plane, extract two-dimensional boundaries from the footprint point cloud using a boundary extraction algorithm, identify inter-floor connection points as inter-floor traffic entrances, plan inter-floor road network topology paths, link inter-floor road networks and topology road networks to obtain a hybrid road network model, merge the external model and the hybrid road network model, output a three-dimensional building information model, and verify the building structure based on vertex smoothness and texture features.

[0028] The model verification module includes: an inter-layer planning unit and a LOD output unit; The inter-floor planning unit is used to project cross-floor components as inter-floor connection points. The cross-floor components include stairs, elevator shafts, and ramps. The inter-floor connection points are added to the topology diagrams of the upper and lower floors respectively to create a high-rise hybrid road network model. The LOD output unit is used to unify the model reference system, associate the nodes and edges in the road network with the geometric elements in the external model, output a three-dimensional building information model, and perform local curvature analysis and texture feature verification on the model.

[0029] A method for managing 3D building models based on modular design includes the following steps: Step S1. Collect point cloud data of building module entities. Based on the geometric and spectral characteristics of the point cloud, separate the auxiliary point cloud from the building point cloud. Extract the view frustum plane from the building point cloud, establish the bounding box space, and divide the point cloud into independent physical components to obtain the building component model. Step S1 includes: Step S11. Use a TLS scanner or oblique photogrammetry system to collect laser point clouds and densely matched point clouds of prefabricated buildings, register and fuse them to obtain a point cloud dataset, uniformly downsample the point cloud dataset using the voxelized mesh method to remove outliers and scanning noise, and identify point cloud features, including: linearity, planarity, scattering, curvature, elevation and point density. Step S12. Use labeled samples to train a random forest classifier, separate the auxiliary point cloud introduced by non-building features, use density clustering algorithm to extract building facades, merge homogeneous areas, calculate the spatial corner points of the plane intersection line, use the intersection judgment algorithm to filter closed spaces, and output the building component model.

[0030] Step S2. Calculate the center coordinates of each building component model. Using the center coordinates as an index, import all building component models into the component library, classify logical components according to building function or volume, and establish the topological connection relationship between components. Step S2 includes: Step S21. Based on vertex constraints and angle error weighted processing of model vertex features, identify model vertices from corner points. The vertices are located on the plane of the model's exterior facade or on the plane intersection line, and the adjacent edges connected to the vertices are perpendicular or parallel. Vertex constraint adjustment is performed through angle error weighting. The center of the outer bounding box of the building component model is calculated using an equal latitude and longitude difference grid to determine the center coordinates of the building component model. Step S22. Establish a spatial index, determine the grid number based on the center coordinates of the component, import a modular component library containing geometric information, attribute information, spatial information and connection relationship information, establish topological connection relationships between components, and maintain associated boundaries when component boundaries change.

[0031] Step S3. Use building components to build a three-dimensional building model, align the component point cloud in the weighted principal direction of the building assembly, scale each component according to the normal direction and center of gravity of the building surface, merge the building edge features, perform texture mapping based on the exterior walls and common contours, reconstruct the model surface, and obtain the building exterior model. Step S3 includes: Step S31. Extract building component models from the component library, calculate the dominant axis of the building as a whole on the horizontal plane according to the set building assembly requirements, and rotate all component models around the center of the dominant axis to align the horizontal axis of the building coordinate system with the weighted principal direction. Step S32. Establish the connection relationship between components. Using the normal of the building floor plan or ground as the reference plane, calculate the vertical distance from the centroid of the point cloud of the aligned components to the reference plane. Compare the ratio of the vertical distance to the input standard floor height and calculate the scaling factor of each component model. Step S33. Extract the building outline texture from the aligned component model. By least squares straight line fitting, fit textures with the same straight line features into a continuous outline. Based on the model unfolding algorithm, calculate the texture coordinates for each outline surface and establish the correspondence between mesh vertices and texture image pixels.

[0032] Step S4. Segment the internal areas of the building, remap the inter-layer features of the internal walls, establish a road network planning model using the distribution data of the intra-layer areas, apply footprint simulation under the constraints of building rules, and establish an intra-layer topological road network; Step S4 includes: Step S41. Extract the vertices of the inner surface triangular mesh from the external model, select the points located on the continuous plane as the initial seed, and classify the points whose normal vector angle and distance to the plane are both less than the threshold into the same region, thus dividing the interior space of the building into different semantic regions. Step S42. Using line fitting and intersection calculation, identify road connections, including doors, internal staircases and openings. Use the center point of the connection as a node and the center line connecting the nodes as an edge to establish the topology map inside each floor and generate a polygonal road network in the plane of each floor.

[0033] Step S5. Project the cross-floor components onto the plane, use the inter-floor connection points as cross-floor traffic entrances, establish an inter-floor road network, link the inter-floor road network and the topological road network to obtain a hybrid road network model, merge the building exterior model and the hybrid road network model, output a three-dimensional building information model, and verify the building structure based on vertex smoothness and texture features.

[0034] Step S5 includes: Step S51. Project cross-floor components as inter-floor connection points. The cross-floor components include stairs, elevator shafts and ramps. Use a boundary extraction algorithm to extract two-dimensional boundaries from the footprint point cloud. Add the inter-floor connection points to the topology maps of the upper and lower floors respectively, and plan the inter-floor road network topology path. Step S52. Unify the model reference system, associate the nodes and edges in the road network with the geometric elements in the external model, output the three-dimensional building information model, and perform local curvature analysis and texture feature verification on the model.

[0035] Example: Collect multi-source point cloud data, extract point cloud features based on random forest, separate the main building, determine intersection and construct bounding box space, locate center coordinates and pose, import modular component library, analyze semantic constraints, unify component main direction, reconstruct model spatial relationship, optimize model edge features, map and register texture library, segment indoor space and generate inter-floor and cross-floor hybrid road network models respectively, merge and verify 3D building model.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0037] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing three-dimensional building models based on modular design, characterized in that, The method includes the following steps: Step S1. Collect point cloud data of building module entities. Based on the geometric and spectral characteristics of the point cloud, separate the auxiliary point cloud from the building point cloud. Extract the view frustum plane from the building point cloud, establish the bounding box space, and divide the point cloud into independent physical components to obtain the building component model. Step S2. Calculate the center coordinates of each building component model. Using the center coordinates as an index, import all building component models into the component library, classify logical components according to building function or volume, and establish the topological connection relationship between components. Step S3. Use building components to build a three-dimensional building model, align the component point cloud in the weighted principal direction of the building assembly, scale each component according to the normal direction and center of gravity of the building surface, merge the building edge features, perform texture mapping based on the exterior walls and common contours, reconstruct the model surface, and obtain the building exterior model. Step S4. Segment the internal areas of the building, remap the inter-layer features of the internal walls, establish a road network planning model using the distribution data of the intra-layer areas, apply footprint simulation under the constraints of building rules, and establish an intra-layer topological road network; Step S5. Project the cross-floor components onto the plane, use the inter-floor connection points as cross-floor traffic entrances, establish an inter-floor road network, link the inter-floor road network and the topological road network to obtain a hybrid road network model, merge the building exterior model and the hybrid road network model, output a three-dimensional building information model, and verify the building structure based on vertex smoothness and texture features.

2. The method for managing three-dimensional building models based on modular design according to claim 1, characterized in that: Step S1 includes: Step S11. Use a TLS scanner or oblique photogrammetry system to collect laser point clouds and densely matched point clouds of prefabricated buildings, register and fuse them to obtain a point cloud dataset, uniformly downsample the point cloud dataset using the voxelized mesh method to remove outliers and scanning noise, and identify point cloud features, including: linearity, planarity, scattering, curvature, elevation and point density. Step S12. Use labeled samples to train a random forest classifier, separate the auxiliary point cloud introduced by non-building features, use density clustering algorithm to extract building facades, merge homogeneous areas, calculate the spatial corner points of the plane intersection line, use the intersection judgment algorithm to filter closed spaces, and output the building component model.

3. The method for managing three-dimensional building models based on modular design according to claim 2, characterized in that: Step S2 includes: Step S21. Based on vertex constraints and angle error weighted processing of model vertex features, identify model vertices from corner points. The vertices are located on the plane of the model's exterior facade or on the plane intersection line, and the adjacent edges connected to the vertices are perpendicular or parallel. Vertex constraint adjustment is performed through angle error weighting. The center of the outer bounding box of the building component model is calculated using an equal latitude and longitude difference grid to determine the center coordinates of the building component model. Step S22. Establish a spatial index, determine the grid number based on the center coordinates of the component, import a modular component library containing geometric information, attribute information, spatial information and connection relationship information, establish topological connection relationships between components, and maintain associated boundaries when component boundaries change.

4. The method for managing three-dimensional building models based on modular design according to claim 3, characterized in that: Step S3 includes: Step S31. Extract building component models from the component library, calculate the dominant axis of the building as a whole on the horizontal plane according to the set building assembly requirements, and rotate all component models around the center of the dominant axis to align the horizontal axis of the building coordinate system with the weighted principal direction. Step S32. Establish the connection relationship between components. Using the normal of the building floor plan or ground as the reference plane, calculate the vertical distance from the centroid of the point cloud of the aligned components to the reference plane. Compare the ratio of the vertical distance to the input standard floor height and calculate the scaling factor of each component model. Step S33. Extract the building outline texture from the aligned component model. By least squares straight line fitting, fit textures with the same straight line features into a continuous outline. Based on the model unfolding algorithm, calculate the texture coordinates for each outline surface and establish the correspondence between mesh vertices and texture image pixels.

5. The method for managing three-dimensional building models based on modular design according to claim 4, characterized in that: Step S4 includes: Step S41. Extract the vertices of the inner surface triangular mesh from the external model, select the points located on the continuous plane as the initial seed, and classify the points whose normal vector angle and distance to the plane are both less than the threshold into the same region, thus dividing the interior space of the building into different semantic regions. Step S42. Using line fitting and intersection calculation, identify road connections, including doors, internal staircases and openings. Use the center point of the connection as a node and the center line connecting the nodes as an edge to establish the topology map inside each floor and generate a polygonal road network in the plane of each floor. Step S5 includes: Step S51. Project cross-floor components as inter-floor connection points. The cross-floor components include stairs, elevator shafts and ramps. Use a boundary extraction algorithm to extract two-dimensional boundaries from the footprint point cloud. Add the inter-floor connection points to the topology maps of the upper and lower floors respectively, and plan the inter-floor road network topology path. Step S52. Unify the model reference system, associate the nodes and edges in the road network with the geometric elements in the external model, output the three-dimensional building information model, and perform local curvature analysis and texture feature verification on the model.

6. A three-dimensional building model management system based on modular design, characterized in that, The system includes the following modules: point cloud forming module, component management module, assembly and merging module, internal expansion module, and model verification module; The point cloud forming module is used to collect point cloud data of building module entities. Based on the random forest algorithm, it classifies the point cloud according to the geometric and spectral features, calculates the features in the neighborhood of each point, separates the auxiliary point cloud from the building point cloud, extracts the point cloud into individual entities through the density clustering extraction module, extracts the view frustum plane from the point cloud, calculates the spatial corner points of the plane intersection line, uses the intersection judgment algorithm to filter the closed space, establishes the bounding box space, and divides the point cloud into independent physical components to obtain the building component model. The component management module is used to process the vertex features of the model based on vertex constraints and angle error weighting, perform vertex constraint adjustment through angle error weighting, calculate the outer bounding box center of the building component model with equal latitude and longitude difference grid, determine the center coordinates of the building component model, import all building component models into the modular component library with the center coordinates as index, divide the logical components in the library according to building function or volume, establish the topological connection relationship between components, and maintain the associated boundary when the component boundary changes. The assembly and merging module is used to build a three-dimensional building model using building components. It aligns the component point cloud in the weighted principal direction of the building assembly constraints, determines the module scaling height based on the normal direction and centroid of the building surface, merges building edge features using line fitting technology, reconstructs the model surface using a triangular mesh reconstruction algorithm, performs texture mapping based on the building model texture library, corrects erroneous planar topology connections, deletes redundant corner points, and obtains the external model. The internal expansion module is used to divide the internal area of ​​the building according to the flatness by using the region growing algorithm, and to stitch and register the texture images by remapping the interlayer features of the internal wall surface and according to the scale-invariant feature transformation of the texture, remove texture noise, and apply footprint simulation under the constraints of building rules by utilizing the distribution of the intralayer area to establish a topological road network. The model verification module is used to project cross-floor components onto a plane, extract two-dimensional boundaries from the footprint point cloud using a boundary extraction algorithm, identify inter-floor connection points as inter-floor traffic entrances, plan inter-floor road network topology paths, link inter-floor road networks and topology road networks to obtain a hybrid road network model, merge the external model and the hybrid road network model, output a three-dimensional building information model, and verify the building structure based on vertex smoothness and texture features.

7. A modular design-based three-dimensional building model management system according to claim 6, characterized in that: The point cloud forming module includes: a data acquisition unit, a point cloud classification unit, and a single-unit extraction unit; The data acquisition unit is used to acquire laser point clouds and densely matched point clouds of prefabricated buildings using a TLS scanner or oblique photogrammetry system, and then register and fuse them to obtain a point cloud dataset. The point cloud classification unit is used to uniformly downsample the point cloud dataset using a voxelized grid method to remove outliers and scanning noise, and to identify point cloud features, including: linearity, planarity, scattering, curvature, elevation, and point density. The single-unit extraction unit is used to train a random forest classifier using labeled samples, separate the auxiliary point cloud introduced by non-building features, extract building facades using density clustering algorithm, merge homogeneous regions, and output building component models.

8. A three-dimensional building model management system based on modular design according to claim 7, characterized in that: The component management module includes: a vertex optimization unit and an indexing unit; The vertex optimization unit is used to identify model vertices from corner points. The vertices are located on the plane of the model's exterior facade or on the intersection of planes, and the adjacent edges connected to the vertices are perpendicular or parallel to each other. The inbound index unit is used to establish a spatial index, determine the grid number by the center coordinates of the component, and import a modular component library containing geometric information, attribute information, spatial information, and connection relationship information.

9. A three-dimensional building model management system based on modular design according to claim 8, characterized in that: The assembly and merging module includes: an intelligent assembly unit, a geometric reconstruction unit, and a texture mapping unit; The intelligent assembly unit is used to extract building component models from the component library, calculate the dominant axis of the building as a whole on the horizontal plane according to the set building assembly requirements, and rotate all component models around the center of the dominant axis to align the horizontal axis of the building coordinate system with the weighted principal direction. The geometric reconstruction unit is used to establish the connection relationship between components. Taking the normal of the building floor plan or ground as the reference plane, it calculates the vertical distance from the centroid of the point cloud of the aligned component to the reference plane, compares the ratio of the vertical distance to the input standard floor height, and calculates the scaling factor of each component model. The texture mapping unit is used to extract the building outline texture from the aligned component model. By least squares straight line fitting, textures with the same straight line features are fitted into continuous contour lines. Based on the model unfolding algorithm, texture coordinates are calculated for each contour surface, and the correspondence between mesh vertices and texture image pixels is established.

10. A three-dimensional building model management system based on modular design according to claim 9, characterized in that: The internal expansion module includes: a region segmentation unit and a road network planning unit; The region segmentation unit is used to extract the vertices of the inner surface triangular mesh from the external model, select points located on the continuous plane as initial seeds, and classify points whose normal vector angle and distance to the plane are both less than the threshold into the same region, thereby dividing the interior space of the building into different semantic regions. The road network planning unit is used to identify road connections, including doors, internal staircases and openings, by using straight line fitting and intersection calculation. The center point of the connection is used as a node, and the center line connecting the nodes is used as an edge, thereby establishing a topology map inside each floor and generating a polygonal road network in the plane of each floor. The model verification module includes: an inter-layer planning unit and a LOD output unit; The inter-floor planning unit is used to project cross-floor components as inter-floor connection points. The cross-floor components include stairs, elevator shafts, and ramps. The inter-floor connection points are added to the topology diagrams of the upper and lower floors respectively to create a high-rise hybrid road network model. The LOD output unit is used to unify the model reference system, associate the nodes and edges in the road network with the geometric elements in the external model, output a three-dimensional building information model, and perform local curvature analysis and texture feature verification on the model.

Images