Building information model generation method, system and device, medium and program product

Abstract

The invention provides a building information model generation method, system and device, a medium and a program product. The generation method comprises the steps of obtaining a building point cloud model; performing multi-plane segmentation on the building point cloud model to obtain a three-dimensional outer wall point cloud and a three-dimensional roof plane point cloud; performing feature extraction on the three-dimensional outer wall point cloud and the three-dimensional roof plane point cloud to obtain outer wall features and roof features; generating a facade drawing based on the three-dimensional outer wall point cloud; and generating a building information model based on the outer wall features, the roof features and the facade drawing. The building information model is generated by fusing unmanned aerial vehicle multi-angle visual collection, multi-plane segmentation and a lightweight deep learning algorithm, full-process automation of existing building surveying and mapping and BIM generation is achieved, an efficient and low-cost digital solution is provided for city updating projects such as old community reconstruction, and the large-scale reconstruction requirement is met.

Description

Technical Field

[0001] The present disclosure relates to the technical field of building information modeling, and in particular to a method, system, device, medium, and program product for generating a building information model. Background Art

[0002] In recent years, the integrated application of BIM technology in the entire life cycle has been accelerated, especially in the field of renovation and upgrading of existing buildings. A certain department has explicitly proposed to build a "new model of real estate development", requiring the use of digital technology to achieve full life cycle management of houses, and focusing on promoting livelihood projects such as urban renewal and renovation of old communities. These indicate that digital surveying and mapping of existing buildings and BIM (Building Information Modeling) modeling have become strategic needs and are an important support for achieving green buildings, smart cities and sustainable development.

[0003] According to statistics, there are currently about 220,000 old residential communities built before 2,000 in the country, involving nearly 39 million households. These existing buildings generally have problems such as missing original drawings and complex structures. Traditional manual surveying methods rely on point-by-point measurement methods using total stations or laser rangefinders, which are time-consuming, labor-intensive, inefficient, and costly, making it difficult to meet the needs of large-scale renovation. Summary of the Invention

[0004] The technical problem to be solved by the present disclosure is to overcome the defects of the traditional manual surveying method in the prior art that relies on the point-by-point measurement method of total station or laser rangefinder, which is time-consuming, labor-intensive, inefficient, costly, and difficult to meet the needs of large-scale renovation, and to provide a method, system, equipment, medium and program product for generating a building information model.

[0005] The present disclosure solves the above technical problems through the following technical solutions:

[0006] A first aspect of the present disclosure provides a method for generating a building information model, the method comprising:

[0007] Obtain a point cloud model of a building;

[0008] Performing multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud;

[0009] Performing feature extraction on the three-dimensional exterior wall point cloud and the three-dimensional roof plane point cloud to obtain exterior wall features and roof features;

[0010] generating a facade drawing based on the three-dimensional exterior wall point cloud;

[0011] A building information model is generated based on the exterior wall features, roof features and the facade drawings.

[0012] Preferably, the step of obtaining a building point cloud model includes:

[0013] Acquire a sequence of building surface images;

[0014] A building point cloud model is generated based on the building surface image sequence.

[0015] Preferably, the step of performing multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud comprises:

[0016] Filtering a vertical plane candidate set from the building point cloud model;

[0017] A statistical filtering algorithm is used to remove outlier noise points in the vertical plane candidate set to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud.

[0018] Preferably, the exterior wall features and roof features include at least one of a building plane polygon outline, a roof boundary outline, and a roof reference height, and the step of extracting features from the three-dimensional exterior wall point cloud and the three-dimensional roof plane point cloud to obtain the exterior wall features and roof features includes:

[0019] Projecting the three-dimensional exterior wall point cloud along the Z axis onto the XOY plane to generate a two-dimensional projection image, and performing edge detection and polygon approximation processing to obtain a plane polygon outline of the building;

[0020] Obtaining a target elevation difference of the three-dimensional exterior wall point cloud as the exterior wall height, and using the exterior wall outline as the roof boundary outline;

[0021] The median elevation value of the three-dimensional roof plane point cloud is obtained as the roof reference height.

[0022] Preferably, the step of generating a facade drawing based on the three-dimensional exterior wall point cloud includes:

[0023] Projecting the three-dimensional exterior wall point cloud along the normal vector direction to generate a vertical orthophoto;

[0024] Performing semantic segmentation on the facade orthophoto to identify door and window areas;

[0025] Performing geometric shape processing on the door and window areas to obtain door and window geometric parameters;

[0026] A facade drawing is generated based on the door and window geometric parameters.

[0027] Preferably, the step of generating a building information model based on the exterior wall features, roof features and the facade drawings comprises:

[0028] Generate exterior wall features based on exterior wall outlines;

[0029] Extracting boundary and elevation data from the exterior wall features, and obtaining roof features based on the boundary and elevation data;

[0030] The exterior wall features and the roof features are inserted into the wall coordinates to obtain a building information model.

[0031] A second aspect of the present disclosure provides a system for generating a building information model, the system comprising:

[0032] An acquisition module, used to obtain a point cloud model of a building;

[0033] A segmentation module, configured to perform multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud;

[0034] A feature extraction module is used to extract features from the three-dimensional exterior wall point cloud and the three-dimensional roof plane point cloud to obtain exterior wall features and roof features;

[0035] A first generating module is configured to generate a facade drawing based on the three-dimensional exterior wall point cloud;

[0036] The second generating module is configured to generate a building information model based on the exterior wall features, roof features and the facade drawings.

[0037] Preferably, the acquisition module includes:

[0038] A first acquisition unit is used to acquire a sequence of building surface images;

[0039] The first generating unit is configured to generate a building point cloud model based on the building surface image sequence.

[0040] Preferably, the segmentation module includes:

[0041] A screening unit, configured to screen out a set of vertical plane candidates from the building point cloud model;

[0042] The removal unit is used to remove outlier noise points in the vertical plane candidate set by using a statistical filtering algorithm to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud.

[0043] Preferably, the exterior wall features and roof features include at least one of a building plane polygon outline, a roof boundary outline, and a roof reference height, and the feature extraction module includes:

[0044] A projection unit is used to project the three-dimensional exterior wall point cloud along the Z axis onto the XOY plane to generate a two-dimensional projection image, and perform edge detection and polygon approximation processing to obtain a plane polygon outline of the building;

[0045] a second acquisition unit, configured to acquire a target elevation difference of the three-dimensional exterior wall point cloud as the exterior wall height, and to use the exterior wall outline as the roof boundary outline;

[0046] The third acquiring unit is configured to acquire a median elevation value of the three-dimensional roof plane point cloud as a roof reference height.

[0047] Preferably, the first generating module includes:

[0048] A second generating unit is configured to project the three-dimensional exterior wall point cloud along a normal vector direction to generate a vertical orthophoto;

[0049] A recognition unit, configured to perform semantic segmentation on the facade orthophoto to identify door and window areas;

[0050] A processing unit, configured to perform geometric shape processing on the door and window areas to obtain door and window geometric parameters;

[0051] The third generating unit is used to generate a facade drawing based on the door and window geometric parameters.

[0052] Preferably, the second generating module includes:

[0053] a fourth generating unit, configured to generate exterior wall features based on the exterior wall contour;

[0054] a fourth acquisition unit, configured to extract boundary and elevation data from the exterior wall features, and acquire roof features based on the boundary and elevation data;

[0055] The fifth acquiring unit is configured to insert the exterior wall features and the roof features into the wall coordinates to obtain a building information model.

[0056] A third aspect of the present disclosure provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and configured to run on the processor, wherein when the processor executes the computer program, the method for generating a building information model as described in the first aspect is implemented.

[0057] A fourth aspect of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method for generating a building information model according to the first aspect is implemented.

[0058] A fifth aspect of the present disclosure provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for generating a building information model as described in the first aspect is implemented.

[0059] On the basis of conforming to the common sense in this field, the above-mentioned preferred conditions can be arbitrarily combined to obtain the preferred embodiments of the present disclosure.

[0060] The positive progress of this disclosure is:

[0061] This paper integrates drone multi-angle visual acquisition, multi-plane segmentation and lightweight deep learning algorithm to generate building information models, realizing the full process automation of existing building surveying and BIM generation, providing efficient and low-cost digital solutions for urban renewal projects such as the renovation of old communities, and meeting the needs of large-scale renovation. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flowchart of the method for generating a building information model provided in Example 1 of the present disclosure.

[0063] Figure 2 Schematic diagram of the BIM model generation process provided for Examples 1 and 2 of the present disclosure.

[0064] Figure 3 Schematic diagram of the elevation drawing generation process provided for Examples 1 and 2 of the present disclosure.

[0065] Figure 4 A comparison chart of the BIM model reconstruction results provided in Examples 1 and 2 of the present disclosure.

[0066] Figure 5 This is a module diagram of a building information model generation system provided in Example 2 of the present disclosure.

[0067] Figure 6 This is a structural diagram of an electronic device for implementing a method for generating a building information model according to embodiment 3 of the present disclosure. DETAILED DESCRIPTION

[0068] The present disclosure is further illustrated below by way of examples, but the present disclosure is not limited to the scope of the examples.

[0069] In the embodiments of the present disclosure, prefixes such as "first" and "second" are used only to distinguish different description objects, and have no limiting effect on the position, order, priority, quantity or content of the described objects. In the embodiments of the present disclosure, the use of prefixes such as ordinal numbers to distinguish description objects does not constitute a limitation on the described objects. For the statement of the described objects, please refer to the description in the context of the claims or embodiments, and no unnecessary limitations should be constituted due to the use of such prefixes. In addition, in the description of this embodiment, unless otherwise specified, the meaning of "plurality" is two or more.

[0070] In the embodiments of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0071] Example 1

[0072] Figure 1 This is a flow chart of a method for generating a building information model provided in Example 1 of the present disclosure, such as Figure 1 As shown, the generation method includes:

[0073] S1. Obtaining a point cloud model of a building;

[0074] S2. Perform multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud;

[0075] In this embodiment, the RANSAC algorithm is used to perform multi-plane segmentation on the building point cloud model, and classification is based on the normal vector direction to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud. Specifically, the RANSAC algorithm is used to perform multi-plane segmentation on the building point cloud model, and a dynamic iteration termination condition is set: it stops when the proportion of points in the plane changes by less than 1% within 50 consecutive iterations, and a normal vector constraint condition is set: the angle threshold between the vertical wall normal vector and the Z axis is 5°.

[0076] S3, extracting features from the 3D exterior wall point cloud and the 3D roof plane point cloud to obtain exterior wall features and roof features;

[0077] S4. Generate facade drawings based on 3D exterior wall point clouds;

[0078] S5. Generate a building information model based on the exterior wall features, roof features and facade drawings.

[0079] In this embodiment, a parametric BIM model is constructed through the Revit API.

[0080] This embodiment integrates point cloud geometry segmentation and deep learning technology. By integrating drone multi-angle visual acquisition, multi-plane segmentation and lightweight deep learning algorithms to generate building information models, it realizes the full process automation of existing building surveying and BIM generation, and provides efficient and low-cost digital solutions for urban renewal projects such as the renovation of old communities, meeting the needs of large-scale renovation.

[0081] In an optional embodiment, S1 includes:

[0082] Acquire a sequence of building surface images;

[0083] Generate a building point cloud model based on a sequence of building surface images.

[0084] In this example, a sequence of building surface images was acquired through multi-angle drone aerial photography, and a high-precision building point cloud model was generated using ContextCapture Center software. For example, a DJI Matrice 300RTK drone equipped with a 45-megapixel Zenmuse P1 camera was used. An automatic flight path was planned using GS Pro software, with a flight altitude of 80 meters, 85% heading overlap, and 80% lateral overlap. A total of 368 images of the building's east, south, west, and north facades, as well as the top, were collected. ContextCapture Center 2023 software was used for 3D reconstruction to generate the building point cloud model. (Specifically, a server cluster integrated with ContextCapture Center software executed the SfM 3D reconstruction algorithm to generate the point cloud model.)

[0085] In an optional embodiment, S2 includes:

[0086] Screening vertical plane candidate sets from the building point cloud model;

[0087] In this embodiment, a vertical plane candidate set is screened from the building point cloud model based on the normal vector constraint condition, and the angle threshold between the normal vector and the vertical direction is set to 5°.

[0088] The statistical filtering algorithm is used to remove outlier noise points in the vertical plane candidate set to obtain the three-dimensional exterior wall point cloud and the three-dimensional roof plane point cloud.

[0089] In an optional embodiment, the exterior wall features and the roof features include at least one of a building plane polygonal outline, a roof boundary outline, and a roof base height, and S3 includes:

[0090] Project the 3D exterior wall point cloud along the Z axis to the XOY plane to generate a 2D projection image, and perform edge detection and polygon approximation processing to obtain the plane polygon outline of the building;

[0091] In this embodiment, the three-dimensional exterior wall point cloud is projected along the Z axis onto the XOY plane to generate a two-dimensional projection image, the Canny edge detection algorithm is used to extract the building outline boundary, and then the Douglas-Peucker algorithm is used to perform polygon approximation to obtain the planar polygonal outline of the building. Specifically, the three-dimensional exterior wall point cloud is projected along the Z axis onto the XOY plane to generate a projection image (resolution 5 cm / pixel), the Canny edge detection (Gaussian filter standard deviation σ=1.5) is applied to extract the building outline boundary, and the Douglas-Peucker algorithm is used to perform polygon simplification (setting the line segment simplification distance threshold to 10 cm) to obtain the planar polygonal outline of the building.

[0092] Obtain the target elevation difference of the 3D exterior wall point cloud as the exterior wall height, and use the exterior wall outline as the roof boundary outline;

[0093] In this embodiment, based on the projected point cloud elevation data, the target elevation difference (e.g., the maximum elevation difference) of each 3D exterior wall point cloud is calculated as the exterior wall height, and the exterior wall outline is used as the roof boundary outline;

[0094] Get the median elevation value of the 3D roof plane point cloud as the roof base height.

[0095] In this embodiment, the outer wall outline is used as the roof boundary outline, a subset of the roof plane point cloud in the three-dimensional roof plane point cloud is screened, and the median value of the three-dimensional roof plane point cloud elevation is calculated as the roof reference height.

[0096] In an optional embodiment, S4 includes:

[0097] Project the 3D exterior wall point cloud along the normal vector direction to generate a vertical orthophoto;

[0098] Perform semantic segmentation on the facade orthophoto to identify door and window areas;

[0099] Perform geometric shape processing on the door and window areas to obtain the door and window geometric parameters;

[0100] In this embodiment, the door and window geometric parameters include center coordinates (x, y), width w, and height h.

[0101] Generate facade drawings based on door and window geometry parameters.

[0102] In this embodiment, each 3D exterior wall point cloud is projected along the normal vector to generate a 5mm / pixel facade orthophoto, and a lightweight convolutional network is used for segmentation. Specifically, the 3D exterior wall point cloud is projected onto a 2D plane along the normal vector direction to generate a facade orthophoto, and the projection resolution is set to 5mm / pixel; a bilinear interpolation algorithm is used to fill the blank areas of the projection in the 2D plane; a lightweight convolutional network is used to perform pixel-level semantic segmentation on the facade orthophoto to identify the door and window areas; it should be noted that the semantic segmentation of the lightweight convolutional encoding-decoding model structure is designed. The segmentation model (e.g., lightweight convolutional network) adopts a 3-layer encoder-decoder structure, and each encoder layer contains 3×3 convolution kernels + batch normalization + ReLU activation; specifically, the 3-layer encoder uses 3×3 convolution kernels with batch normalization layers (i.e., 3×3 convolution kernels, stride=2), and the 3-layer decoder uses 2×2 deconvolution kernels with skip connections. Input layer: 1024×1024 pixel blocks (stride 512 pixels); output layer: single-channel classification results (wall, window, door); CUDA accelerated processing: 12FPS inference speed is achieved on NVIDIA GTX 1080Ti (i.e., 1024×1024 pixel blocks (stride 512 pixels) are used for large-size images, and real-time segmentation is achieved through CUDA parallel acceleration). In addition, the output layer implements multi-category classification through the Sigmoid function, i.e., the output layer uses a single-channel Sigmoid activation and realizes three types of pixel classification of wall (0), window (1), and door (2) through a threshold of 0.5.

[0103] Furthermore, the contour detection and minimum enclosing rectangle fitting methods are used to simplify the geometric shape of the door and window areas (that is, the door and window areas are morphologically optimized, and the door and window geometric parameters are extracted by minimum enclosing rectangle fitting); facade drawings are automatically generated based on the door and window geometric parameters and the AutoCAD secondary development interface; specifically, the pywin32 library is used to drive AutoCAD to automatically generate facade drawings, including opening drawings, creating door and window graphic elements, and saving the final drawing files (for example, a drawing generator developed based on the pywin32 interface automatically generates building facade drawings through the AutoCAD secondary development interface).

[0104] In an optional embodiment, S5 includes:

[0105] Generate exterior wall features based on exterior wall outlines;

[0106] In this embodiment, the Wall.Create method is called to generate exterior wall features according to the exterior wall outline;

[0107] Extract boundary and elevation data from exterior wall features, and obtain roof features based on boundary and elevation data;

[0108] In this embodiment, the boundary and elevation data of the exterior wall feature are extracted by applying the NewExtrusionRoof method, and the roof feature is created based on the boundary and elevation data;

[0109] The exterior wall features and the roof features are inserted into the wall coordinates to obtain a building information model.

[0110] In this embodiment, FamilyInstance.ByGeometry is used to insert the door and window family instances into the specified coordinates of the wall to obtain the building information model.

[0111] In the specific implementation process, Figure 2 Generate a schematic diagram of the BIM model process, Figure 3 Schematic diagram of the process of generating elevation drawings, Figure 4 This is a comparison chart of BIM model reconstruction results. Specifically, a DJI Matrice 300 RTK drone equipped with a high-precision camera captured multi-angle aerial photography, capturing high-definition image sequences of the building facades and rooftops. Using ContextCapture Center software, 3D point cloud reconstruction was performed using the Structure from Motion (SfM) algorithm to generate a high-precision building point cloud model. The RANSAC algorithm was then used to perform multi-plane segmentation on the building point cloud model. Normal constraints (a 5° angle threshold between the vertical wall normal and the Z-axis) and a dynamic iteration termination mechanism (a change in the in-plane point ratio of less than 1% within 50 consecutive iterations) were introduced to accurately separate exterior wall and roof features.

[0112] The segmented 3D exterior wall point cloud was projected along the Z axis of the building's local coordinate system onto the XOY plane to generate a 2D projection. Building outline polygons were extracted using Canny edge detection (Gaussian filter σ = 1.5) and the Douglas-Peucker algorithm (simplified threshold 10 cm). The exterior wall height was determined by calculating the maximum elevation difference of the projected point cloud. The outline polygon was then used as the roof boundary. A subset of roof plane point clouds was selected from the 3D roof plane point cloud, and the median elevation value was calculated as the roof reference height. For building facade feature extraction, the 3D exterior wall point cloud was projected along the normal vector to generate a 5 mm / pixel resolution facade orthophoto. A lightweight encoder-decoder convolutional network was then applied for pixel-level semantic segmentation. This network utilizes a three-layer encoder architecture (3×3 convolution kernels + batch normalization) and a three-layer decoder architecture (2×2 deconvolution + skip connections). It processes large-scale images using a sliding window strategy (1024×1024 pixel blocks with a stride of 512 pixels) and outputs a single-channel classification result (wall, window, door).

[0113] Based on the semantic segmentation results, minimum bounding rectangle fitting was used to extract door and window geometry. The pywin32 library was used to drive AutoCAD to automatically generate elevation drawings. Finally, wall dimensions, roof parameters, and door and window positioning data (i.e., elevation drawings) were integrated to construct a parametric BIM model using the Revit API. This involved calling the Wall.Create method to generate exterior wall features (e.g., parametric walls), applying NewExtrusionRoof to create roof features, and using FamilyInstance.ByGeometry to precisely locate door and window family instances to create the building information model.

[0114] This implementation integrates drone multi-angle visual acquisition, RANSAC multi-plane segmentation, and lightweight deep learning algorithms to automate the entire process of surveying and mapping existing buildings and generating BIM. This implementation deeply aligns with the strategic needs of "smart construction," providing efficient and low-cost digital solutions for urban renewal projects such as the renovation of older residential communities, with significant socioeconomic benefits and industry-wide adoption.

[0115] Example 2

[0116] Corresponding to the aforementioned embodiment of a method for generating a building information model, the present disclosure also provides an embodiment of a system for generating a building information model.

[0117] Figure 5 This is a module diagram of a building information model generation system provided in Example 2 of the present disclosure, such as Figure 5 As shown, the generation system includes:

[0118] An acquisition module 21 is used to acquire a point cloud model of a building;

[0119] The segmentation module 22 is used to perform multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud;

[0120] In this embodiment, the RANSAC algorithm is used to perform multi-plane segmentation on the building point cloud model, and classification is based on the normal vector direction to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud. Specifically, the RANSAC algorithm is used to perform multi-plane segmentation on the building point cloud model, and a dynamic iteration termination condition is set: it stops when the proportion of points in the plane changes by less than 1% within 50 consecutive iterations, and a normal vector constraint condition is set: the angle threshold between the vertical wall normal vector and the Z axis is 5°.

[0121] A feature extraction module 23 is used to extract features from the 3D exterior wall point cloud and the 3D roof plane point cloud to obtain exterior wall features and roof features;

[0122] A first generating module 24 is configured to generate a facade drawing based on the 3D exterior wall point cloud;

[0123] The second generating module 25 is configured to generate a building information model based on the exterior wall features, roof features and facade drawings.

[0124] In this embodiment, a parametric BIM model is constructed through the Revit API.

[0125] This embodiment integrates point cloud geometry segmentation and deep learning technology. By integrating drone multi-angle visual acquisition, multi-plane segmentation and lightweight deep learning algorithms to generate building information models, it realizes the full process automation of existing building surveying and BIM generation, and provides efficient and low-cost digital solutions for urban renewal projects such as the renovation of old communities, meeting the needs of large-scale renovation.

[0126] In an optional embodiment, the acquisition module includes:

[0127] A first acquisition unit is used to acquire a sequence of building surface images;

[0128] The first generating unit is configured to generate a building point cloud model based on a building surface image sequence.

[0129] In this example, a sequence of building surface images was acquired through multi-angle drone aerial photography, and a high-precision building point cloud model was generated using ContextCapture Center software. For example, a DJI Matrice 300RTK drone equipped with a 45-megapixel Zenmuse P1 camera was used. An automatic flight path was planned using GS Pro software, with a flight altitude of 80 meters, 85% heading overlap, and 80% lateral overlap. A total of 368 images of the building's east, south, west, and north facades, as well as the top, were collected. ContextCapture Center 2023 software was used for 3D reconstruction to generate the building point cloud model. (Specifically, a server cluster integrated with ContextCapture Center software executed the SfM 3D reconstruction algorithm to generate the point cloud model.)

[0130] In an optional embodiment, the segmentation module includes:

[0131] A screening unit, used for screening a vertical plane candidate set from the building point cloud model;

[0132] In this embodiment, a vertical plane candidate set is screened from the building point cloud model based on the normal vector constraint condition, and the angle threshold between the normal vector and the vertical direction is set to 5°.

[0133] The removal unit is used to remove outlier noise points in the vertical plane candidate set using a statistical filtering algorithm to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud.

[0134] In an optional embodiment, the exterior wall features and the roof features include at least one of a building plane polygon outline, a roof boundary outline, and a roof base height, and the feature extraction module includes:

[0135] The projection unit is used to project the three-dimensional exterior wall point cloud along the Z axis to the XOY plane to generate a two-dimensional projection image, and perform edge detection and polygon approximation processing to obtain the plane polygon outline of the building;

[0136] In this embodiment, the three-dimensional exterior wall point cloud is projected along the Z axis onto the XOY plane to generate a two-dimensional projection image, the Canny edge detection algorithm is used to extract the building outline boundary, and then the Douglas-Peucker algorithm is used to perform polygon approximation to obtain the planar polygonal outline of the building. Specifically, the three-dimensional exterior wall point cloud is projected along the Z axis onto the XOY plane to generate a projection image (resolution 5 cm / pixel), the Canny edge detection (Gaussian filter standard deviation σ=1.5) is applied to extract the building outline boundary, and the Douglas-Peucker algorithm is used to perform polygon simplification (setting the line segment simplification distance threshold to 10 cm) to obtain the planar polygonal outline of the building.

[0137] The second acquisition unit is used to obtain the target elevation difference of the three-dimensional exterior wall point cloud as the exterior wall height, and use the exterior wall outline as the roof boundary outline;

[0138] In this embodiment, based on the projected point cloud elevation data, the target elevation difference (e.g., the maximum elevation difference) of each 3D exterior wall point cloud is calculated as the exterior wall height, and the exterior wall outline is used as the roof boundary outline;

[0139] The third acquisition unit is used to obtain the median elevation value of the three-dimensional roof plane point cloud as the roof reference height.

[0140] In this embodiment, the outer wall outline is used as the roof boundary outline, a subset of the roof plane point cloud in the three-dimensional roof plane point cloud is screened, and the median value of the three-dimensional roof plane point cloud elevation is calculated as the roof reference height.

[0141] In an optional embodiment, the first generating module includes:

[0142] The second generating unit is used to project the three-dimensional exterior wall point cloud along the normal vector direction to generate a vertical orthophoto;

[0143] The recognition unit is used to perform semantic segmentation on the facade orthophoto and identify the door and window areas;

[0144] A processing unit, used for performing geometric shape processing on the door and window areas to obtain geometric parameters of the doors and windows;

[0145] In this embodiment, the door and window geometric parameters include center coordinates (x, y), width w, and height h.

[0146] The third generating unit is used to generate a facade drawing based on the geometric parameters of the doors and windows.

[0147] In this embodiment, each 3D exterior wall point cloud is projected along the normal vector to generate a 5mm / pixel facade orthophoto, and a lightweight convolutional network is used for segmentation. Specifically, the 3D exterior wall point cloud is projected onto a 2D plane along the normal vector direction to generate a facade orthophoto, and the projection resolution is set to 5mm / pixel; a bilinear interpolation algorithm is used to fill the blank areas of the projection in the 2D plane; a lightweight convolutional network is used to perform pixel-level semantic segmentation on the facade orthophoto to identify the door and window areas; it should be noted that the semantic segmentation of the lightweight convolutional encoding-decoding model structure is designed. The segmentation model (e.g., lightweight convolutional network) adopts a 3-layer encoder-decoder structure, and each encoder layer contains 3×3 convolution kernels + batch normalization + ReLU activation; specifically, the 3-layer encoder uses 3×3 convolution kernels with batch normalization layers (i.e., 3×3 convolution kernels, stride=2), and the 3-layer decoder uses 2×2 deconvolution kernels with skip connections. Input layer: 1024×1024 pixel blocks (stride 512 pixels); output layer: single-channel classification results (wall, window, door); CUDA accelerated processing: 12FPS inference speed is achieved on NVIDIA GTX 1080Ti (i.e., 1024×1024 pixel blocks (stride 512 pixels) are used for large-size images, and real-time segmentation is achieved through CUDA parallel acceleration). In addition, the output layer implements multi-category classification through the Sigmoid function, i.e., the output layer uses a single-channel Sigmoid activation and realizes three types of pixel classification of wall (0), window (1), and door (2) through a threshold of 0.5.

[0148] Furthermore, the contour detection and minimum enclosing rectangle fitting methods are used to simplify the geometric shape of the door and window areas (that is, the door and window areas are morphologically optimized, and the door and window geometric parameters are extracted by minimum enclosing rectangle fitting); facade drawings are automatically generated based on the door and window geometric parameters and the AutoCAD secondary development interface; specifically, the pywin32 library is used to drive AutoCAD to automatically generate facade drawings, including opening drawings, creating door and window graphic elements, and saving the final drawing files (for example, a drawing generator developed based on the pywin32 interface automatically generates building facade drawings through the AutoCAD secondary development interface).

[0149] In an optional embodiment, the second generating module includes:

[0150] a fourth generating unit, configured to generate exterior wall features based on the exterior wall contour;

[0151] In this embodiment, the Wall.Create method is called to generate exterior wall features according to the exterior wall outline;

[0152] a fourth acquisition unit, configured to extract boundary and elevation data from the exterior wall features, and acquire roof features based on the boundary and elevation data;

[0153] In this embodiment, the boundary and elevation data of the exterior wall feature are extracted by applying the NewExtrusionRoof method, and the roof feature is created based on the boundary and elevation data;

[0154] The fifth acquisition unit is used to insert the exterior wall features and the roof features into the wall coordinates to obtain a building information model.

[0155] In this embodiment, FamilyInstance.ByGeometry is used to insert the door and window family instances into the specified coordinates of the wall to obtain the building information model.

[0156] In the specific implementation process, Figure 2 Generate a schematic diagram of the BIM model process, Figure 3 Schematic diagram of the process of generating elevation drawings, Figure 4 This is a comparison chart of BIM model reconstruction results. Specifically, a DJI Matrice 300 RTK drone equipped with a high-precision camera captured multi-angle aerial photography, capturing high-definition image sequences of the building facades and rooftops. Using ContextCapture Center software, 3D point cloud reconstruction was performed using the Structure from Motion (SfM) algorithm to generate a high-precision building point cloud model. The RANSAC algorithm was then used to perform multi-plane segmentation on the building point cloud model. Normal constraints (a 5° angle threshold between the vertical wall normal and the Z-axis) and a dynamic iteration termination mechanism (a change in the in-plane point ratio of less than 1% within 50 consecutive iterations) were introduced to accurately separate exterior wall and roof features.

[0157] The segmented 3D exterior wall point cloud was projected along the Z axis of the building's local coordinate system onto the XOY plane to generate a 2D projection. Building outline polygons were extracted using Canny edge detection (Gaussian filter σ = 1.5) and the Douglas-Peucker algorithm (simplified threshold 10 cm). The exterior wall height was determined by calculating the maximum elevation difference of the projected point cloud. The outline polygon was then used as the roof boundary. A subset of roof plane point clouds was selected from the 3D roof plane point cloud, and the median elevation value was calculated as the roof reference height. For building facade feature extraction, the 3D exterior wall point cloud was projected along the normal vector to generate a 5 mm / pixel resolution facade orthophoto. A lightweight encoder-decoder convolutional network was then applied for pixel-level semantic segmentation. This network utilizes a three-layer encoder architecture (3×3 convolution kernels + batch normalization) and a three-layer decoder architecture (2×2 deconvolution + skip connections). It processes large-scale images using a sliding window strategy (1024×1024 pixel blocks with a stride of 512 pixels) and outputs a single-channel classification result (wall, window, door).

[0158] Based on the semantic segmentation results, minimum bounding rectangle fitting was used to extract door and window geometry. The pywin32 library was used to drive AutoCAD to automatically generate elevation drawings. Finally, wall dimensions, roof parameters, and door and window positioning data (i.e., elevation drawings) were integrated to construct a parametric BIM model using the Revit API. This involved calling the Wall.Create method to generate exterior wall features (e.g., parametric walls), applying NewExtrusionRoof to create roof features, and using FamilyInstance.ByGeometry to precisely locate door and window family instances to create the building information model.

[0159] This implementation integrates drone multi-angle visual acquisition, RANSAC multi-plane segmentation, and lightweight deep learning algorithms to automate the entire process of surveying and mapping existing buildings and generating BIM. This implementation deeply aligns with the strategic needs of "smart construction," providing efficient and low-cost digital solutions for urban renewal projects such as the renovation of older residential communities, with significant socioeconomic benefits and industry-wide adoption.

[0160] Since the system embodiments generally correspond to the method embodiments, reference will be made to the description of the method embodiments for relevant details. The system embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components of the units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the disclosed solution.

[0161] Example 3

[0162] Figure 6 This is a structural diagram of an electronic device shown in Example 3 of the present disclosure. The electronic device includes a memory, a processor, and a computer program stored in the memory and used to run on the processor. When the processor executes the computer program, it implements the method for generating a building information model described in any of the above embodiments. Figure 6 The electronic device 90 shown is only an example and should not limit the functionality and scope of use of the embodiments of the present disclosure.

[0163] like Figure 6 As shown, the electronic device 90 may be a general-purpose computing device, such as a server device. Components of the electronic device 90 may include, but are not limited to, the at least one processor 91, the at least one memory 92, and a bus 93 connecting different system components (including the memory 92 and the processor 91).

[0164] The bus 93 includes a data bus, an address bus, and a control bus.

[0165] The memory 92 may include a volatile memory, such as a random access memory (RAM) 921 and / or a cache memory 922 , and may further include a read-only memory (ROM) 923 .

[0166] The memory 92 may also include a program tool 925 (or utility) having a set (at least one) of program modules 924, such program modules 924 including but not limited to: an operating system, one or more application programs, other program modules and program data, each of which or some combination may include an implementation of a network environment.

[0167] The processor 91 executes various functional applications and data processing by running the computer programs stored in the memory 92, such as the method for generating a building information model provided in any of the above embodiments.

[0168] The electronic device 90 can also communicate with one or more external devices 94 (e.g., keyboards, pointing devices, etc.). Such communication can be performed through an input / output (I / O) interface 95. In addition, the electronic device 90 can also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 96. Figure 6 As shown, the network adapter 96 communicates with other modules of the electronic device 90 via the bus 93. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 90, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (RAID) systems, tape drives, and data backup storage systems.

[0169] It should be noted that although several units / modules or sub-units / modules of the electronic device are mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.

[0170] Example 4

[0171] Embodiment 4 of the present disclosure further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for generating a building information model provided in any of the above embodiments.

[0172] The readable storage medium may include, but is not limited to, a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0173] Example 5

[0174] Embodiment 5 of the present disclosure further provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-mentioned methods for generating a building information model.

[0175] The program code for executing the computer program product of the present disclosure may be written in any combination of one or more programming languages, and the program code may be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on the remote device.

[0176] While specific embodiments of the present disclosure have been described above, those skilled in the art will appreciate that these are merely illustrative and that the scope of protection of the present disclosure is defined by the appended claims. Those skilled in the art may make various changes or modifications to these embodiments without departing from the principles and essence of the present disclosure, and such changes and modifications are intended to fall within the scope of protection of the present disclosure.

Claims

1. A method for generating a building information model, characterized in that: The generation method comprises: Obtain a point cloud model of a building; Performing multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud; Performing feature extraction on the three-dimensional exterior wall point cloud and the three-dimensional roof plane point cloud to obtain exterior wall features and roof features; generating a facade drawing based on the three-dimensional exterior wall point cloud; A building information model is generated based on the exterior wall features, roof features and the facade drawings.

2. The method for generating a building information model according to claim 1, wherein: The step of obtaining a building point cloud model comprises: Acquire a sequence of building surface images; A building point cloud model is generated based on the building surface image sequence.

3. The method for generating a building information model according to claim 1, wherein: The step of performing multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud comprises: Filtering a vertical plane candidate set from the building point cloud model; A statistical filtering algorithm is used to remove outlier noise points in the vertical plane candidate set to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud.

4. The method for generating a building information model according to claim 1, wherein: The exterior wall features and roof features include at least one of a building plane polygon outline, a roof boundary outline, and a roof reference height. The step of extracting features from the three-dimensional exterior wall point cloud and the three-dimensional roof plane point cloud to obtain the exterior wall features and roof features includes: Projecting the three-dimensional exterior wall point cloud along the Z axis onto the XOY plane to generate a two-dimensional projection image, and performing edge detection and polygon approximation processing to obtain a plane polygon outline of the building; Obtaining a target elevation difference of the three-dimensional exterior wall point cloud as the exterior wall height, and using the exterior wall outline as the roof boundary outline; The median elevation value of the three-dimensional roof plane point cloud is obtained as the roof reference height.

5. The method for generating a building information model according to claim 1, wherein: The step of generating a facade drawing based on the three-dimensional exterior wall point cloud includes: Projecting the three-dimensional exterior wall point cloud along the normal vector direction to generate a vertical orthophoto; Performing semantic segmentation on the facade orthophoto to identify door and window areas; Performing geometric shape processing on the door and window areas to obtain door and window geometric parameters; A facade drawing is generated based on the door and window geometric parameters.

6. The method for generating a building information model according to claim 1, wherein: The step of generating a building information model based on the exterior wall features, roof features and the facade drawings comprises: Generate exterior wall features based on exterior wall outlines; Extracting boundary and elevation data from the exterior wall features, and obtaining roof features based on the boundary and elevation data; The exterior wall features and the roof features are inserted into the wall coordinates to obtain a building information model.

7. A building information model generation system, characterized in that: The generation system comprises: An acquisition module is used to obtain a point cloud model of a building; A segmentation module, configured to perform multi-plane segmentation on the building point cloud model to obtain a three-dimensional exterior wall point cloud and a three-dimensional roof plane point cloud; A feature extraction module is used to extract features from the three-dimensional exterior wall point cloud and the three-dimensional roof plane point cloud to obtain exterior wall features and roof features; A first generating module is configured to generate a facade drawing based on the three-dimensional exterior wall point cloud; The second generating module is configured to generate a building information model based on the exterior wall features, roof features and the facade drawings.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and configured to run on the processor, wherein: When the processor executes the computer program, the method for generating a building information model according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for generating a building information model according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for generating a building information model according to any one of claims 1 to 6 is implemented.

Images