BIM model generation method and device based on three-dimensional point cloud structure

Abstract

The invention discloses a BIM (Building Information Modeling) generation method and device based on a three-dimensional point cloud structure. The method comprises the following steps: acquiring a point cloud model of the structure by using a three-dimensional laser scanner; segmenting each component in the point cloud model; according to IFC specifications, constructing a geometric model based on the surface and a damage model based on a triangulation network; based on the IFC attribute set, geometric parameters and damage parameters of the structure are recorded; and automatically outputting a BIM file in an IFC format on the basis of an Ifopen Shell library, generating a BIM model with damage, and realizing structured storage of a structural geometric model and damage information. The defect that damage data obtained by a traditional structure model and daily inspection cannot be stored at the same time is overcome, two tasks of modeling and damage extraction based on point cloud data are achieved, visualization of the structure and damage is achieved in a BIM mode, and management and maintenance of existing structure data are facilitated.

Description

Technical Field

[0001] This invention relates to the field of BIM application technology, specifically to the field of BIM application technology in structural inspection, and particularly to a method and apparatus for generating BIM models of structures based on three-dimensional point clouds. Background Technology

[0002] Regular structural inspections are a crucial part of structural lifecycle management and a vital guarantee for ensuring the safe operation of structures. However, currently, regular structural inspections rely primarily on manual labor, and inspection results are mostly stored in document and image formats. This leads to problems such as disorganized historical inspection information and difficulties in tracing and identifying the source of problems.

[0003] In recent years, BIM (Building Information Modeling) has gradually entered the field of structural inspection and is playing an increasingly important role, especially in structural inspection and recording. BIM can simultaneously store structural model information, inspection information, structural damage information, etc. Compared with the traditional document-based recording method, BIM can more effectively organize and store relevant structural information.

[0004] Meanwhile, with the development of 3D laser scanning technology, the geometric models of existing structures in BIM can now be obtained based on point cloud models. Furthermore, research on point cloud-based surface damage identification is increasing, making point clouds crucial data for simultaneously obtaining structural geometric and damage information. Therefore, automatically generating BIM models based on 3D point clouds will facilitate rapid, periodic structural inspections, systematically store structural and inspection results, and visualize relevant structural information. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in current structural inspection and data storage methods by providing a method and apparatus for generating BIM models of structures based on 3D point clouds. This invention develops a method including a deep learning-based point cloud segmentation algorithm, a point cloud-based 3D modeling algorithm, a point cloud-based damage identification and quantification algorithm, an IFC attribute set construction method, and an automated IFC file output method. It primarily solves the problems of low efficiency in existing structural inspections, ineffective storage of inspected data, and difficulty in visualization.

[0006] The objective of this invention is achieved through the following technical solution: The first aspect of this invention provides a method for generating a BIM model based on a three-dimensional point cloud structure, comprising the following steps: (1) Obtaining the point cloud model of the structure: The structure is scanned from all directions using a 3D laser scanner mounted on a drone to obtain a 3D point cloud, and the 3D point cloud is processed to obtain the point cloud model of the structure. (2) Point cloud segmentation: The PointNet point cloud segmentation neural network is used to automatically segment each component in the point cloud model of the structure; (3) IFC geometric model: The various components of the structure are modeled by fitting and stretching to obtain the IFC geometric model; (4) IFC damage model: The damage identification algorithm based on point cloud color and curvature is used to identify surface damage in the point cloud; the Delaunay triangulation algorithm is used to construct a triangular mesh of the point cloud of the damaged area, establish the IFC damage model, and perform quantization. (5) IFC attribute set: Create corresponding IFC attribute sets according to the structure type and damage type respectively; (6) Automated output of IFC files: Based on the IFC geometric model, IFC damage model and its IPC attribute set, a programming language is used to automatically output IFC format BIM files.

[0007] Furthermore, the process of processing the 3D point cloud to obtain the point cloud model of the structure specifically includes: The ICP point cloud registration algorithm is used to register the 3D point cloud to the global coordinate system. Then, outliers in the point cloud are filtered out and downsampling is performed to obtain the point cloud model of the structure.

[0008] Furthermore, the PointNet point cloud segmentation neural network needs to be trained before use. When training the PointNet point cloud segmentation neural network, a corresponding training set needs to be established, which consists of point cloud models obtained from actual scanning and artificially generated point cloud models.

[0009] Furthermore, in step (3), modeling each component of the structure needs to be done in the local coordinate system of each component, and the geometric models are all based on IFC representation.

[0010] Furthermore, the damage recognition algorithm based on point cloud color and curvature is specifically implemented through the following method: First, the color of the point cloud is converted to grayscale, and the curvature of the point cloud is estimated; then, the grayscale of the point cloud is adaptively filtered, that is, every point in the point cloud is traversed. And use the KNN algorithm to search for its nearest neighbor set within radius r. Then use Gaussian uniform grayscale replacement The grayscale value is then used; subsequently, each point is iterated over again. And search for its nearest neighbor set within radius r. ,like grayscale less than The average grayscale value, and If the curvature at a point is greater than a preset curvature threshold, then a judgment is made. The point cloud represents the possible damaged areas. After completing the above process, the DBSCAN clustering algorithm is used to cluster the possible points, filter out the few discrete points, and finally obtain the actual damaged area point cloud cluster.

[0011] Furthermore, the IFC damage model in step (4) is based on the Delaunay triangulation of the IFC specification, and the quantification of damage is also based on the triangulation.

[0012] Furthermore, the IFC attribute set in step (5) can be customized, and the IFC attribute set is linked to the corresponding IFC entity.

[0013] Furthermore, in step (6), the automated output IFC file is implemented based on the IfcOpenShell library. The IFC format BIM file is supported by BIM software and can be visualized in BIM software.

[0014] A second aspect of the present invention provides a BIM model generation apparatus based on a three-dimensional point cloud structure, comprising one or more processors and a memory, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the above-described BIM model generation method based on a three-dimensional point cloud structure.

[0015] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, is used to implement the above-described method for generating a BIM model of a structure based on a three-dimensional point cloud.

[0016] The beneficial effects of this invention are as follows: This invention automatically performs point cloud segmentation, 3D point cloud modeling, and point cloud-based damage identification and quantification based on 3D point clouds. Furthermore, this invention adopts the IFC standard to establish attribute sets for geometric and damage models, links and records their respective attributes, and finally, based on the IfcOpenShell library, this invention achieves automated output of IFC format BIM files. This invention realizes the process from point cloud to BIM, facilitating the structured storage and management of geometric models, damage, and attribute information. At the same time, the final output BIM file also realizes information visualization, greatly facilitating structural inspection and management. Attached Figure Description

[0017] Figure 1 This is a flowchart of the BIM model generation method based on three-dimensional point cloud structure according to the present invention; Figure 2This is the IFC structure diagram in the BIM model of the present invention, which includes the IFC geometric model, the IFC damage model, and the IFC attribute set, as well as the relationships between them; Figure 3 This is a schematic diagram of the IFC attribute set of the present invention; Figure 4 This is a code example diagram of the IFC damage model based on Delaunay triangulation representation of the present invention; Figure 5 This is a schematic diagram of a BIM model generation device based on a three-dimensional point cloud structure according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0019] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0020] It should be understood that although the terms first, second, third, etc., may be used in this invention to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of this invention, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0021] The present invention will now be described in detail with reference to the accompanying drawings. Unless otherwise specified, the features of the following embodiments and implementations can be combined with each other.

[0022] See Figure 1 The BIM model generation method based on three-dimensional point cloud structure of the present invention specifically includes the following steps: (1) Obtaining the point cloud model of the structure: The structure is scanned from all directions using a 3D laser scanner mounted on a drone to obtain a 3D point cloud, and the 3D point cloud is processed to obtain the point cloud model of the structure.

[0023] Further, the point cloud model of the structure is obtained by processing the 3D point cloud. Specifically, the 3D point cloud is registered to the global coordinate system using the ICP point cloud registration algorithm. Then, outliers in the point cloud are filtered out and downsampling is performed to obtain the point cloud model of the structure.

[0024] Specifically, a drone equipped with a 3D laser scanner is used. After calibration, the drone's flight path is planned in detail, allowing the drone to perform a 3D laser scanner to scan the structure from all directions. The resulting 3D point cloud is then registered to the global coordinate system using the ICP point cloud registration algorithm. Afterward, outliers in the point cloud are filtered out, and downsampling is performed to obtain a basic point cloud model of the structure.

[0025] Furthermore, when filtering outliers in the point cloud, either radius-based or statistical filtering methods can be used. The radius-based method filters out points with a k-neighborhood radius greater than r; the statistical method filters out points where the average distance from the point of interest to its k-neighborhood is no longer within a given range. Afterward, downsampling can be easily performed using a voxel-based downsampling method.

[0026] (2) Point cloud segmentation: The PointNet point cloud segmentation neural network is used to automatically segment each component in the point cloud model of the structure.

[0027] Furthermore, the PointNet point cloud segmentation neural network needs to be trained before it can be used. When training the PointNet point cloud segmentation neural network, a corresponding training set needs to be established. This training set can be composed of point cloud models obtained from actual scanning and artificially generated point cloud models.

[0028] Specifically, the PointNet point cloud segmentation neural network is used to automatically segment the point cloud model of the structure. The input to this PointNet network is the point cloud model of the structure, and the output is the segmented point cloud model of each component of the structure. Before use, the PointNet network needs to be trained, requiring the creation of a corresponding training set. To address the issue of insufficient 3D point cloud training data in displays, the training set is partly obtained from real scans and partly generated automatically based on programming languages ​​and mathematical models of the structure, thus enriching and enhancing the dataset. The training process of the PointNet network is common in this field and will not be elaborated upon here.

[0029] (3) IFC (Industrial Foundation) geometric model: The various components of the structure are modeled by fitting, stretching and other methods to obtain the IFC geometric model.

[0030] Furthermore, modeling each component of the structure requires working in the local coordinate system of each component, and the geometric models are all based on IFC representation.

[0031] Specifically, for each component segmented in step (2), a 3D model is performed based on the characteristics of the component. First, each component needs to be transformed into its own local coordinate system. The choice of local coordinate system should make the modeling simpler, and the component should be upright in the local coordinate system as much as possible. For components such as beams and columns in the structure, their cross-sections can be fitted first, and then stretched along a certain path, and represented by IfcExtrudedAreaSolid; for components composed of multiple planes, the planes of the component can be separated first and fitted by the least squares method, and then represented by IfcFacetedBrep. These IFC geometric models are finally represented by IfcShapeRepresentation, which can be represented as a BIM model. When modeling each component, the relationships between the relevant points, lines, surfaces, and geometric bodies are as follows: Figure 2 As shown.

[0032] It should be understood that IfcExtrudedAreaSolid is a basic entity in the IFC standard used to define 3D geometry. It is a three-dimensional block generated by "stretching" a 2D cross-section along a certain distance in one direction. IfcFacetedBrep is an important way in the IFC standard to represent 3D solid geometry. It defines a solid by describing its boundary, specifically, a closed shell composed of a series of flat polygonal facets. IfcShapeRepresentation is a core and fundamental concept in the IFC standard. It directly relates to how geometric information is organized and expressed. It is a container used to illustrate how a building component can be depicted or visualized geometrically.

[0033] (4) IFC damage model: The damage identification algorithm based on point cloud color and curvature is used to identify the surface damage of the structure in the point cloud; the Delaunay triangulation algorithm is used to construct the triangular network of the point cloud of the damaged area, establish the IFC damage model, and perform quantization.

[0034] In this embodiment, damage is identified using the color and curvature features of point clouds. Generally, at the site of point cloud damage, the color of the point cloud differs from other areas, and the grayscale is darker; simultaneously, the surface of the damaged area is uneven, with significant curvature variations. These two features are fundamental characteristics for determining damage. Therefore, the damage identification algorithm based on point cloud color and curvature is implemented using the following method: First, the point cloud is converted to grayscale, and its curvature is estimated. Then, the grayscale of the point cloud is adaptively filtered, that is, every point in the point cloud is traversed. And use the KNN algorithm to search for its nearest neighbor set within radius r. Then use Gaussian uniform grayscale replacement The grayscale value. Then, the process is repeated for each point. And search for its nearest neighbor set within radius r. ,like grayscale less than The average grayscale value, and If the curvature at a point is greater than a certain curvature threshold, then it is determined that... The point cloud represents the possible damaged areas. After completing the above process, the DBSCAN clustering algorithm is used to cluster the possible points, filtering out the few discrete points, and finally obtaining the actual damaged area point cloud clusters.

[0035] It should be noted that this damage identification algorithm is implemented by comparing the grayscale value of the point of interest with the grayscale values ​​of its neighbors and setting a point cloud curvature threshold. When using this damage identification algorithm to identify potential damage, a clustering algorithm is still needed to filter out incorrectly identified points.

[0036] Furthermore, surface damage can be modeled using IfcTriangulatedFaceSet and IfcSurfaceFeature. This requires first constructing a Delaunay triangulation of the point cloud clusters in the damaged area, and then using IfcRelAggregates to link the triangulation to the corresponding solid model.

[0037] Furthermore, during quantization, the size of the damaged area can be calculated based on a triangular mesh. For common crack damage, it is also necessary to combine the α-shape algorithm to extract the crack edge lines, and then use the orthogonal skeleton line method to calculate the length and width of the crack.

[0038] (5) IFC attribute set: Create corresponding IFC attribute sets according to the structure type and damage type respectively.

[0039] Furthermore, IFC attribute sets can be customized, and the IFC attribute sets are linked to the corresponding IFC entities.

[0040] Specifically, the property set is represented using IfcPropertySet, and the property is represented using IfcProperty. For example... Figure 3 As shown, some basic attribute information is listed. For the IFC geometric model, two attribute sets are defined: one is the structural state attribute set, such as whether the structure is damaged, the number of damages, etc.; the other is the geometric dimension attribute set, such as the length, width, height of the geometric body, and structural deformation, displacement, etc. For the IFC damage model, a damage information attribute set is defined, mainly including damage type, damage location, damage degree, etc. The attribute set can use standard attribute sets, or it can be customized according to actual needs. Finally, IfcRelDefinedByProperties is used to link the IFC attribute set to the corresponding IFC entity, completing the binding between the model and the attributes.

[0041] (6) Automated output of IFC files: Based on the IFC geometric model, IFC damage model and its IPC attribute set, a programming language is used to automatically output IFC format BIM files.

[0042] Furthermore, the automated output of IFC files is implemented based on the IfcOpenShell library. IFC format BIM files are supported by BIM software and can be visualized within the software.

[0043] Specifically, the IfcOpenShell library is a Python library for quickly parsing and outputting IFC file formats. After obtaining the IFC geometric model, IFC damage model, and their attribute information, this invention, based on the IfcOpenShell library and the Python programming language, automatically outputs IFC format files programmatically. Figure 4 The image shows an automated STEP file, which represents IFC surface damage based on a triangular mesh.

[0044] In summary, this invention proposes an automated method for creating BIM models of structures based on 3D point clouds, including geometric entity models, damage models, and their attribute sets. This invention realizes the structural storage and corresponding visualization of structural model information and attribute information, improves the organization and storage methods of traditional structural data, enhances information management efficiency, and facilitates the automation of structural inspection.

[0045] Corresponding to the aforementioned embodiments of the BIM model generation method based on three-dimensional point cloud structures, the present invention also provides embodiments of a BIM model generation device based on three-dimensional point cloud structures.

[0046] See Figure 5 The present invention provides a BIM model generation device based on a three-dimensional point cloud structure, comprising one or more processors and a memory, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the BIM model generation method based on a three-dimensional point cloud structure in the above embodiment.

[0047] The embodiment of the BIM model generation device based on the structure of three-dimensional point clouds of this invention can be applied to any device with data processing capabilities, such as a computer. The device embodiment can be implemented through software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor of any data processing device loading the corresponding computer program instructions from non-volatile memory into memory and executing them. From a hardware perspective, such as... Figure 5 The diagram shown is a hardware structure diagram of any device with data processing capabilities, including the BIM model generation device based on a 3D point cloud structure according to the present invention. (Except for...) Figure 5 In addition to the processor, memory, network interface, and non-volatile memory shown, any data processing device in the embodiment may also include other hardware depending on the actual function of the data processing device, which will not be described in detail here.

[0048] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0049] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0050] This invention also provides a computer-readable storage medium storing a program thereon, which, when executed by a processor, implements the BIM model generation method based on a three-dimensional point cloud structure as described in the above embodiments.

[0051] The computer-readable storage medium can be an internal storage unit of any data processing device described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be any data processing device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0052] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating a BIM model based on a three-dimensional point cloud-based structure, characterized by, The method comprises the following steps: (1) obtaining a point cloud model of a structure: using a three-dimensional laser scanner carried by a UAV to scan the structure at various orientations to obtain a three-dimensional point cloud, and obtaining a point cloud model of the structure after processing the three-dimensional point cloud; (2) point cloud segmentation: automatically segmenting each component in the point cloud model of the structure by using a pointnet point cloud segmentation neural network; (3) IFC geometric model: modeling each component of the structure by using fitting and stretching to obtain an IFC geometric model; (4) IFC damage model: identifying surface damage of the structure in the point cloud by using a damage identification algorithm based on point cloud color and curvature; constructing a triangular mesh of the point cloud of the damage area by using a Delaunay triangulation algorithm, establishing an IFC damage model, and quantifying; (5) IFC attribute set: creating corresponding IFC attribute sets according to the structure type and the damage type; (6) automatically outputting an IFC file: based on the IFC geometric model, the IFC damage model, and the IFC attribute set, automatically outputting an IFC format BIM file by using a programming language. 2.The BIM model generation method based on a three-dimensional point cloud structure according to claim 1, characterized in that, The point cloud model of the structure is obtained by processing the three-dimensional point cloud, specifically comprising: The three-dimensional point cloud is registered into a global coordinate system by using an ICP point cloud registration algorithm, then the outliers in the point cloud are filtered out, and the sampling is reduced, so that the point cloud model of the structure is obtained. 3.The BIM model generation method based on a three-dimensional point cloud structure according to claim 1, characterized in that, The pointnet point cloud segmentation neural network needs to be trained before use. When training the pointnet point cloud segmentation neural network, a corresponding training set needs to be established, which is composed of point cloud models obtained by actual scanning and artificially generated point cloud models. 4.The method of claim 1, wherein, In step (3), the modeling of each component of the structure needs to be performed in the local coordinate system of each component, and the geometric model is based on IFC representation. 5.The BIM model generation method based on a three-dimensional point cloud structure according to claim 1, characterized in that, The damage identification algorithm based on point cloud color and curvature is implemented by the following method: First, the color of the point cloud is grayed, and the curvature of the point cloud is estimated; then the gray of the point cloud is adaptively filtered, that is, each point in the point cloud is traversed , and the KNN algorithm is used to search for the neighbor point set within its radius r , and then the gray of is replaced by the Gaussian uniform gray of ; Then, iterate through each point again. And search for its nearest neighbor set within radius r. ,like Grayscale less than The average gray level, and If the curvature at a point is greater than a preset curvature threshold, then a judgment is made. The point cloud represents the possible damaged areas. After completing the above process, the DBSCAN clustering algorithm is used to cluster the possible points, filter out the few discrete points, and finally obtain the actual damaged area point cloud cluster. 6.The BIM model generation method based on a three-dimensional point cloud structure according to claim 1, characterized in that, The IFC damage model in step (4) is a Delaunay triangular mesh based on IFC specification, and the quantification of damage is also based on the triangular mesh. 7.The BIM model generation method based on a three-dimensional point cloud structure according to claim 1, characterized in that, The IFC attribute set in step (5) can be customized, and the IFC attribute set is linked to the corresponding IFC entity. 8.The BIM model generation method based on a three-dimensional point cloud structure according to claim 1, characterized in that, In step (6), the automatic output of the IFC file is realized based on the IfcOpenShell library, the IFC format BIM file is supported by BIM software, and visualization can be realized in BIM software.

9. A BIM model generation apparatus based on a three-dimensional point cloud structure, comprising one or more processors and a memory, characterized in that, The memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to realize the BIM model generation method of the structure based on the three-dimensional point cloud in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, A program is stored thereon, which is executed by a processor to realize the BIM model generation method of the structure based on the three-dimensional point cloud in any one of claims 1-8.

Images