BIM building model construction method and system for building construction

Abstract

The invention relates to the technical field of building design, in particular to a BIM building model construction method and system.The BIM building model construction method comprises the steps that firstly, based on the plane continuity characteristic of a wall body structure in a building construction environment, rough segmentation is conducted on point cloud through an RANSAC multi-plane fitting algorithm, and a wall area and outer points of a complex structure are preliminarily separated; further, for complex structure areas such as curved surfaces and arc connection areas, model contribution factors are dynamically generated by calculating relative distances and geometric coherence between outer points and a fitting plane, and higher weights are given to point clouds in a high-curvature area; according to the method, the registration process is optimized in combination with the weighted ICP algorithm, and the key structure area is aligned preferentially, so that the damage to the integrity and continuity of a complex structure is reduced, the registered building point cloud data after registration is more accurate, and the accuracy of the BIM model constructed according to the registered building point cloud data is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of point cloud data processing, and in particular to a BIM building model construction method and system for building construction. Background Art

[0002] In construction management, point cloud data can help designers better understand the relationship between the construction site and the surrounding environment. Within the BIM model, designers can combine point cloud data with architectural design plans for visual analysis, thereby optimizing the building layout. BIM models and point cloud data can also be used to perform construction collision detection, identifying potential collisions in advance and preventing construction accidents. Therefore, building point cloud data collected from multiple angles is typically registered and fused to improve data consistency and integrity. Existing technologies typically utilize the ICP point cloud registration algorithm to register building point cloud data, and construct BIM models based on the registered building point cloud data.

[0003] However, due to the dust and equipment vibration in the construction environment, a large number of noise points will be introduced. The spatial positions of these points are randomly distributed and may be mistakenly judged as the nearest points. At the same time, for areas with large curvature changes, such as curved surfaces and complex geometric structures (such as arcs and hollow components), the nearest points may come from different curved surfaces (such as the junction of adjacent walls and beams). At this time, directly using the ICP point cloud registration algorithm for point cloud registration may destroy the integrity and continuity of some complex structures, making the accuracy of the registered building point cloud data low, thereby affecting the accuracy of the constructed BIM model. Summary of the Invention

[0004] In order to solve the technical problem that when directly using the ICP point cloud registration algorithm for point cloud registration, the integrity and continuity of some structures may be destroyed, resulting in low accuracy of the registered building point cloud data, the purpose of this application is to provide a BIM building model construction method and system for building construction. The technical solutions adopted are as follows: In a first aspect, the present application provides a BIM building model construction method for building construction, comprising: Scanning the building with at least two radars to obtain initial building point cloud data; determining an initial wall area in the initial building point cloud data based on a plane fitting method; wherein data points outside the initial wall area are off-wall points; Based on the extension of similar normal vectors of the wall points located at the edge of each initial wall area, all corresponding local wall edges are determined; based on the similarity of the distance distribution between each external wall point and all local wall edges, each external wall point structure set is determined; based on the position of each external wall point structure set relative to each local wall edge, a plane fitting is performed to determine the corresponding external wall point fitting plane; According to the complexity of the neighborhood structure of each point cloud in the off-wall point structure set, a model contribution factor of each point cloud is determined; and point cloud cleaning is performed according to the model contribution factor to obtain registered building point cloud data.

[0005] Furthermore, the process of obtaining the local edge of the wall includes: Obtaining an initial wall edge of each initial wall area; screening out false edge points based on a standard deviation of normal vectors corresponding to wall points on the initial wall edge in the initial wall area; and obtaining a reference wall edge after screening out the false edge points from the initial wall edge. The invention relates to a method for determining a wall point in the reference wall edge as an edge determination point; connecting adjacent edge determination points in sequence in the order of the reference wall edge to determine a determination edge line representing the line between each edge determination point and the next edge determination point; using the angle between the determination edge line of each edge determination point and the determination edge line of the previous edge determination point as the determination angle of each edge determination point; removing the determination edge line of each edge determination point when the determination angle of the edge determination point is greater than a preset first determination threshold, or when the difference between the determination angle of the corresponding edge determination point and the determination angle of the previous edge determination point is greater than a preset second determination threshold; and dividing the reference wall edge into at least two local wall edges after traversing all edge determination points.

[0006] Furthermore, the process of obtaining the false edge points includes: In each initial wall area, the corresponding reference normal vector is determined based on the mean of the normal vectors of all other wall points outside the initial wall edge; the angle between the normal vector of each wall point in the initial wall edge and the reference normal vector is used as the corresponding edge judgment angle; and the wall point whose corresponding edge judgment angle is less than a preset angle threshold is regarded as a false edge point.

[0007] Furthermore, the process of obtaining the wall external point structure set includes: A clustering space is constructed with the minimum Euclidean distance between an off-wall point and each local edge of the wall as each dimension; a clustering coordinate point of each off-wall point in the clustering space is determined according to the minimum Euclidean distance between each off-wall point and each local edge of the wall; a k-means clustering analysis is performed on all clustering coordinate points to obtain at least two clusters; and a set corresponding to all off-wall points corresponding to each clustering coordinate point in each cluster is used as a set of off-wall point structures.

[0008] Furthermore, the process of obtaining the wall external point fitting plane includes: Determine the reference relative distance of each local edge of the wall surface based on the average distance between all external wall points in each external wall point structure set and each local edge of the wall surface; determine the boundary position distance based on the overall change trend of the reference relative distance corresponding to each external wall point structure set; and arrange all reference relative distances of each external wall point structure set in ascending order to obtain a reference relative distance sequence; When the number of reference relative distances before the demarcation position distance is less than a preset number threshold, the local wall edges corresponding to the demarcation position distance in the reference relative distance sequence and all the local wall edges corresponding to all the reference relative distances before the demarcation position distance are used as the fitted wall edges of the corresponding external wall point structure set; when the number of reference relative distances before the demarcation position distance is greater than or equal to the preset number threshold, all the local wall edges corresponding to all the reference relative distances before the demarcation position distance in the reference relative distance sequence are used as the fitted wall edges of the corresponding external wall point structure set; According to each wall external point structure set and all corresponding fitted wall edges, plane fitting is performed using the RANSAC point cloud multi-plane fitting segmentation algorithm to determine the wall external point fitting plane corresponding to each wall external point structure set.

[0009] Furthermore, the process of obtaining the boundary position distance includes: In the reference relative distance sequence, the difference between each reference relative distance and the previous reference relative distance is used as the distance span value of each reference relative distance; and the reference relative distance with the largest distance span value in the reference relative distance sequence is used as the boundary position distance.

[0010] Furthermore, the process of obtaining the model contribution factor includes: In the wall point structure set of each wall point fitting plane, all point cloud data points are traversed by the KNN nearest neighbor algorithm to obtain respective fitting model sequences; each point cloud data point in each fitting model sequence is sequentially used as a target data point; the fitting model sequence in which the target data point is located is used as a target data point model; a model is constructed based on the target data point model to obtain a target fitting model; the target data point in the fitting model sequence in which the target data point is located is removed and the model is constructed to obtain a comparative fitting model; Determine the corresponding reference contribution degree based on the position distribution of each point cloud data point in the target point cloud sequence relative to the target fitting model and the wall point fitting plane; determine the model fitting degree of the target fitting model based on the average of the reference contribution degrees of all point cloud data points in the target point cloud sequence; Based on the principle of obtaining the model fitting degree of the target point cloud sequence, the model fitting degree of the comparison fitting model is calculated; the difference between the model fitting degree of the target fitting model and the model fitting degree of the comparison fitting model is normalized to determine the model contribution factor of the target data point.

[0011] Furthermore, the process of obtaining the reference contribution degree includes: In the target point cloud sequence, the minimum Euclidean distance from each point cloud data point to the target fitting model is used as the reference matching accuracy of each point cloud data point; the negative correlation mapping value of the minimum Euclidean distance from each point cloud data point to the wall fitting plane corresponding to the target point cloud sequence is used as the curvature complexity weight of each point cloud data point; and the corresponding reference contribution degree is determined according to the product between the reference matching accuracy and the curvature complexity weight.

[0012] Furthermore, the process of obtaining the registered building point cloud data includes: Obtain a minimization error function corresponding to the singular value decomposition method in ICP point cloud registration; introduce the model contribution factor of each point cloud in each off-wall point cloud structure set as a weight into the minimization error function to determine a weighted minimization error function; continue ICP point cloud registration according to the weighted minimization error function to obtain registered building point cloud data.

[0013] In a second aspect, the present application provides a BIM building model construction system for building construction, the system comprising: A data acquisition and preprocessing module is configured to obtain initial building point cloud data by scanning the building using at least two radars; determine an initial wall area in the initial building point cloud data based on a plane fitting method; wherein data points outside the initial wall area are off-wall points; The module for determining the external wall point fitting plane is used to determine all corresponding local wall edges based on the extension of similar normal vectors of wall points located at the edge of each initial wall area; determine each external wall point structure set based on the similarity of the distance distribution between each external wall point and all local wall edges; and perform plane fitting based on the position of each external wall point structure set relative to each local wall edge to determine the corresponding external wall point fitting plane. The building point cloud data construction module is used to determine the model contribution factor of each point cloud according to the complexity of the neighborhood structure of each point cloud in the wall point structure set; and perform point cloud cleaning according to the model contribution factor to obtain aligned building point cloud data.

[0014] In a third aspect, the present application provides a computer device comprising a memory and a processor. The memory is configured to store computer program code, and the processor is configured to call and execute the computer program code from the memory to perform the method of the first aspect or any embodiment of the first aspect of the present application.

[0015] In a fourth aspect, the present application provides a computer program product, comprising a computer program code. When the computer program code is executed, the method of the first aspect or any embodiment of the first aspect of the present application is performed.

[0016] In a fifth aspect, the present application provides a computer-readable storage medium, which stores computer program code. When the computer program code is executed, it performs the method of the first aspect of the present application or any embodiment of the first aspect.

[0017] This application has the following beneficial effects: This application first uses the RANSAC multi-plane fitting algorithm to roughly segment the point cloud based on the planar continuity characteristics of the main wall structure in the construction environment, and preliminarily separates the wall area and the external points of the complex structure; further, for complex structure areas such as curved surfaces and arc-shaped connection areas, by calculating the relative distance and geometric continuity between the external points and the fitting plane, the model contribution factor is dynamically generated, and the point cloud in the high curvature area is given a higher weight; combined with the weighted ICP algorithm to optimize the alignment process, the key structural areas are aligned first, thereby reducing the damage to the integrity and continuity of the complex structure, making the registered building point cloud data more accurate after alignment, and improving the accuracy of the BIM model constructed based on the registered building point cloud data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A flowchart of a method for constructing a BIM building model for building construction provided by one embodiment of the present invention; Figure 2 A structural diagram of a BIM building model construction system for building construction provided by one embodiment of the present invention; Figure 3 The present invention provides a schematic diagram of a computer device structure according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the following, in combination with the accompanying drawings and preferred embodiments, a BIM building model construction method and system for building construction proposed by the present invention, its specific implementation method, structure, characteristics and effects are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment, and the specific features, structures or characteristics in one or more embodiments may be combined in any suitable form. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0021] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0022] The following describes in detail a method and system for constructing a BIM building model for building construction provided by the present invention in conjunction with the accompanying drawings.

[0023] This application embodiment provides a BIM building model construction method for building construction, please refer to Figure 1 , which shows a flow chart of a method for constructing a BIM building model for building construction provided by one embodiment of the present invention, the method comprising: Step S101: obtaining initial building point cloud data by scanning the building with at least two radars; determining an initial wall area in the initial building point cloud data based on a plane fitting method; wherein data points outside the initial wall area are off-wall points.

[0024] In a specific implementation of an embodiment of the present invention, at each radar deployment point, the radar is controlled to remain level using a spirit level and then fixed on a tripod to scan the building to be modeled. The point cloud data collected by each radar deployment point is integrated into a point cloud space to obtain the initial building point cloud data required by this application; wherein the point cloud spacing is set to be less than or equal to 10 mm.

[0025] The conventional IPC iterative closest point method only considers the distance between point cloud data in space as the basis for determining redundant points. However, in the current scenario, due to the angular deviation between different deployment points and the complex structure of some building surfaces, some key points may be judged as redundant simply because of their close distance. In this case, it is necessary to extract the structural features of the point cloud based on the possible structural information of the current building surface and perform a rough registration before fine-tuning it using algorithms such as ICP to improve the accurate alignment between point clouds.

[0026] Based on the purpose of preliminary coarse segmentation processing, this application first screens out wall areas with larger areas and simpler structures. In a specific implementation of an embodiment of the present invention, the initial wall area in the initial building point cloud data is identified by the RANSAC point cloud multi-plane fitting segmentation algorithm; wherein, the distance threshold of the RANSAC point cloud multi-plane fitting segmentation algorithm is set to 20mm, which is twice the point cloud accuracy of 10mm, which can accommodate some minor and negligible problems in the construction process. The local deformation of the point cloud caused by the uneven plastering of the wall; the number of iterations is set to 500 times; it should be noted that the RANSAC point cloud multi-plane fitting segmentation algorithm is a technical means well known to those skilled in the art, and the specific parameter settings can be adjusted according to the specific implementation environment, and no further limitation or elaboration is made here.

[0027] After determining each initial wall area, several point cloud data involved in the fitting in each initial wall area are taken as wall points, and the point cloud data that are not judged to be outside the initial wall area of other point cloud data sets are taken as off-wall points for subsequent analysis and processing.

[0028] Step S102: Determine all corresponding local wall edges based on the extension of similar normal vectors of wall points located at the edge of each initial wall area; determine each external wall point structure set based on the similarity of the distance distribution between each external wall point and all local wall edges; perform plane fitting based on the position of each external wall point structure set relative to each local wall edge to determine the corresponding external wall point fitting plane.

[0029] Most of the difficult-to-identify or complex areas in the building structure are external wall points. However, some external wall points may represent some structural features in the wall area. At the same time, the partial structural features composed of these point clouds are usually the connected parts of a relatively complete wall, and there are certain regularities such as curves or locally continuous similar distributions. Therefore, it is necessary to further determine the fitting plane that may also belong to the wall structure based on the local structural features composed of each external wall point.

[0030] Before this, we first need to consider that the initial wall area is obtained by coarse segmentation, so the wall points located at the edge are very easy to be identified as wall points due to their close distance. Therefore, it is necessary to further screen and analyze the edge points to make the final building model more accurate.

[0031] Preferably, in some possible implementations of the embodiments of the present invention, the process of obtaining the local edge of the wall includes: Obtain the initial wall edge of each initial wall region; filter out false edge points based on the standard deviation of the normal vectors corresponding to the wall points in the initial wall region; specifically, in each initial wall region, determine the corresponding reference normal vector based on the mean of the normal vectors of all wall points outside the initial wall edge; use the angle between the normal vector of each wall point in the initial wall edge and the reference normal vector as the corresponding edge determination angle; and define wall points whose corresponding edge determination angles are less than a preset angle threshold as false edge points. Further filtering out the false edge points in the initial wall edge yields a reference wall edge.

[0032] First, for a standard wall, the normal vectors of each point cloud data point tend to be consistent. Therefore, based on the directional deviation of each edge wall point from the reference normal vector, incorrectly identified edge points located on the edge can be filtered out. In one specific implementation of this embodiment of the present invention, the preset angle threshold is set to 10 degrees, which can be adjusted according to the specific implementation environment.

[0033] A wall point in the reference wall edge is used as an edge determination point; adjacent edge determination points are sequentially connected in the order of the reference wall edge to determine a determination edge line representing the line between each edge determination point and the next edge determination point; the angle between the determination edge line of each edge determination point and the determination edge line of the previous edge determination point is used as the determination angle of each edge determination point; when the determination angle of the edge determination point is greater than a preset first determination threshold, or the difference between the determination angle of the corresponding edge determination point and the determination angle of the previous edge determination point is greater than a preset second determination threshold, the determination edge line of each edge determination point is removed; and after traversing all edge determination points, the reference wall edge is divided into at least two local wall edges.

[0034] First, for off-wall points, if the corresponding off-wall point set is a connection part of a relatively complete wall, then these off-wall points are usually closer to one or some local edges of the wall area. For example, for two wall areas with an overall rectangular shape, if there is a fitting plane of the wall connection part between the two wall areas, then the points in the fitting plane will be relatively close to the local edges corresponding to the sides adjacent to the fitting plane in the two rectangles corresponding to the two wall areas; Based on this feature, the present application first divides the edges of the wall area to determine the local edges of each wall; Combined with the characteristics of the polygonal wall, the change of the same local edge is usually relatively stable. Therefore, in order to determine the local edges of the wall more accurately, the present application divides the edge lines by the angle between the connecting lines of adjacent edge points and the direction change between the connecting lines, so that the local edges of each wall obtained are more accurate. In a specific implementation of an embodiment of the present invention, the first judgment threshold is preset to 5 degrees, and the second judgment threshold is preset to 2 degrees, which can be adjusted according to the specific implementation environment. It should be noted that the difference represents the absolute value of the difference, and when the corresponding determination angle or the determination angle of the previous edge determination point cannot be performed due to the absence of an edge point before the edge determination point, the corresponding edge line is not removed, making the embodiment of the present invention more complete.

[0035] For the structure that essentially belongs to the connecting part of the wall, the distance distribution between the external points of the wall and the local edges of each wall is similar. In order to further determine the external points of the structure that truly belongs to the connecting part of the wall, the structure set of each external point of the wall is further determined based on the similarity of the distance distribution between each external point of the wall and all the local edges of the wall; so that the structure set of external point of the wall can truly represent the plane structure corresponding to the connecting part of the wall.

[0036] Preferably, in some possible implementations of the embodiments of the present invention, the process of obtaining the wall external point structure set includes: A clustering space is constructed with the minimum Euclidean distance between an off-wall point and each local edge of the wall as each dimension; the clustering coordinate point of each off-wall point in the clustering space is determined based on the minimum Euclidean distance between each off-wall point and each local edge of the wall; k-means clustering analysis is performed on all clustering coordinate points to obtain at least two clusters; the set corresponding to all off-wall points corresponding to each clustering coordinate point in each cluster is used as the off-wall point structure set. In a specific implementation of an embodiment of the present invention, the optimal number of clusters of k-means clustering is determined by the silhouette coefficient, and the process of determining the optimal number of clusters of k-means clustering by the silhouette coefficient is a technical means well known to those skilled in the art and will not be further elaborated here.

[0037] For any external wall point, the shortest distance to each local wall edge is calculated to determine the relative spatial position of that external point to the overall building structure. Each external wall point has relative distances to multiple wall edges. In the complex structure of connected parts of a wall, most of these distances, excluding those within the wall, are located at intersections, such as corners. Among all the edge distances from each external wall point to all local wall edges, only a few adjacent edge distances are extremely small compared to the remaining edge distances. Therefore, clustering based on the distance to each local wall edge as a multi-dimensional method yields several clusters of external points located between the same wall edges, representing the set of external wall point structures corresponding to each cluster.

[0038] After determining each set of wall-external point structures, the fitting plane corresponding to each set of wall-external point structures can be determined based on the relative position of the set of wall-external point structures relative to each local edge of the wall surface. Preferably, in some possible implementations of the embodiments of the present invention, the process of obtaining the wall-external point fitting plane includes: The reference relative distance of each local edge of the wall is determined based on the average distance between all external wall points of each external wall point structure set and each local edge of the wall surface. The boundary position distance is determined based on the overall change trend of the reference relative distances corresponding to each external wall point structure set. Specifically, all reference relative distances of each external wall point structure set are arranged in ascending order to obtain a reference relative distance sequence. In the reference relative distance sequence, the difference between each reference relative distance and the previous reference relative distance is used as the distance span value of each reference relative distance. The reference relative distance with the largest distance span value in the reference relative distance sequence is used as the boundary position distance.

[0039] First, for the off-wall point structure set that represents the off-wall point fitting plane, its corresponding plane is usually the connected part of the wall. Therefore, the corresponding off-wall point structure set must be relatively close to the local edge of the wall adjacent to the corresponding off-wall point fitting plane, and significantly smaller than other local edges of the wall; and the boundary position distance represents the point where the reference relative distance changes the most. Therefore, the decomposed position distance usually represents the boundary position between the local edge of the wall adjacent to the fitting plane of the off-wall point structure set and the local edge of the wall of the non-adjacent wall. Therefore, the decomposed position distance is further used as the boundary position to determine the local edge of the wall adjacent to the fitting plane of the off-wall point structure set.

[0040] Taking into account the possibility that there is a contingency that causes the number of reference relative distances before the demarcation position distance to be too small, thereby affecting the accuracy of the final obtained wall-outside point fitting plane; therefore, further, when the number of reference relative distances before the demarcation position distance is less than a preset number threshold, the local wall edge corresponding to the demarcation position distance in the reference relative distance sequence and all the local wall edges corresponding to all the reference relative distances before the demarcation position distance are used as the fitting wall edges of the corresponding wall-outside point structure set; when the number of reference relative distances before the demarcation position distance is greater than or equal to the preset number threshold, all the local wall edges corresponding to all the reference relative distances before the demarcation position distance in the reference relative distance sequence are used as the fitting wall edges of the corresponding wall-outside point structure set. In a specific implementation of an embodiment of the present invention, the preset number threshold is set to 2, so that the number of fitting wall edges is at least two, which is consistent with the situation in the implementation scenario where the wall-outside point fitting plane corresponding to the connecting part of the wall is adjacent to at least two wall local edges.

[0041] Based on each set of external wall point structures and all corresponding fitted wall edges, a plane fitting is performed using the RANSAC point cloud multi-plane fitting segmentation algorithm to determine the external wall point fitting plane corresponding to each external wall point structure set. It should be noted that the parameters of the RANSAC point cloud multi-plane fitting segmentation algorithm used here are the same and can be adjusted according to the specific implementation environment. Further explanation is not provided here.

[0042] Step S103: Determine the model contribution factor of each point cloud according to the complexity of the neighborhood structure of each point cloud in the external wall point structure set; perform point cloud cleaning according to the model contribution factor to obtain registered building point cloud data.

[0043] The determination of the off-wall point fitting plane is based on the combination of different local wall edges with similar distances. Since they are not the same wall, there are spatial misalignments. In this case, complex structures are usually located within the area corresponding to the off-wall point fitting edge. The more complex the location of the corresponding point cloud, the more accurate the ICP registration needs to be in the subsequent process to relax the corresponding accuracy requirements to a certain extent to avoid model distortion and overfitting noise.

[0044] Preferably, in some possible implementations of the embodiments of the present invention, the process of obtaining the model contribution factor includes: In the set of off-wall point structures of each off-wall point fitting plane, all point cloud data points are traversed by the KNN nearest neighbor algorithm to obtain respective fitting model sequences; in a specific implementation of an embodiment of the present invention, the range of the KNN nearest neighbor is set to 10 mm; considering that the accuracy of the point cloud data points in this application is 10 mm, even if there is a certain overlap of the point cloud data points between the radar deployment points, it will not exceed 10 mm, so this application sets the range of the KNN nearest neighbor to 10 mm for judgment.

[0045] Each point cloud data point in each fitting model sequence is taken as a target data point in turn; a model is constructed according to the fitting model sequence where the target data point is located to obtain a target fitting model; the target data point in the fitting model sequence where the target data point is located is removed and then a model is constructed to obtain a comparative fitting model; according to the position distribution of each point cloud data point in the target point cloud sequence relative to the target fitting model and the wall point fitting plane, the corresponding reference contribution degree is determined. Specifically, in the target point cloud sequence, the minimum Euclidean distance of each point cloud data point to the target fitting model is used as the reference matching accuracy of each point cloud data point; the negative correlation mapping value of the minimum Euclidean distance of each point cloud data point to the wall point fitting plane corresponding to the target point cloud sequence is used as the curvature complexity weight of each point cloud data point; the corresponding reference contribution degree is determined according to the product between the reference matching accuracy and the curvature complexity weight; the model fitting degree of the target fitting model is determined according to the average of the reference contribution degrees of all point cloud data points in the target point cloud sequence.

[0046] In the process of obtaining the degree of model fit, the greater the distance between the point cloud data point and the corresponding wall point fitting plane, the more likely the point cloud data point is located at a location with greater curvature in the structure. Its participation in the construction of the corresponding fitting model will increase the fitting deviation of the resulting fitting model. Therefore, a negative correlation mapping is performed to obtain a curvature complexity weight to weaken the computational weight of the point cloud data point corresponding to the complex structure location. The smaller the spatial distance between the point cloud data point and the fitted model, the higher the matching accuracy and the better the fitting effect. Therefore, the curvature complexity weight of each point cloud data point in the fitting model and the reference matching accuracy are further combined to characterize the overall degree of model fit of the corresponding fitting model. The greater the model fit, the better the model fit effect of the corresponding fitting model. For the target data point, if the model fit effect of the fitting model deteriorates after the corresponding target data point is removed, and the greater the degree of deterioration, the more important the target data point is, that is, the greater the contribution of the target data point to the construction of the fitting model.

[0047] Therefore, based on the principle of obtaining the model fitting degree of the target point cloud sequence, the model fitting degree of the comparison fitting model is further calculated; that is, after the target fitting model in the process of obtaining the model fitting degree of the target fitting model is replaced with the comparison fitting model, the model fitting degree of the comparison fitting model is obtained. The smaller the model fitting degree of the comparison fitting degree is compared with the model fitting degree of the target fitting model, the higher the contribution of the target data point in the target point cloud sequence, and the more likely the corresponding target data point is to be a sampling point of the complex structure surface between the real wall surfaces. Therefore, the difference between the model fitting degree of the target fitting model and the model fitting degree of the comparison fitting model is further normalized to determine the model contribution factor of the target data point, so that when the model contribution factor is larger, the corresponding point cloud data point is more likely to represent the complex structure surface between the real wall surfaces, and the higher the accuracy required when aligning the point cloud data points.

[0048] In a specific implementation of the embodiment of the present invention, the process of obtaining the model fitting degree of the target fitting model is expressed by the formula: ;in, The target data point The model fit of the corresponding target fitting model; The target data point The number of point cloud data points in the fitted model sequence; The target data point The first point in the target point cloud sequence The minimum Euclidean distance from a point cloud data point to the target fitting model, that is, the corresponding reference matching accuracy; The target data point The first point in the target point cloud sequence The minimum Euclidean distance from a point cloud data point to the corresponding wall outside point fitting plane; is a linear normalization function; The target data point The first point in the target point cloud sequence The curvature complexity weight of each point cloud data point; For each comparison fitting model, the target fitting model's model fit degree corresponding to the target point cloud data point is obtained by replacing the target fitting model with the comparison fitting model in the formulas corresponding to the process of obtaining the target fitting model's model fit degree, and replacing the target point cloud sequence with the fitting model sequence of the comparison fitting model. When normalizing the difference between the model fit degrees of the target fitting model and the comparison fitting model, linear normalization is used, which can be adjusted according to the specific implementation environment and is not further explained here.

[0049] After further determining the model contribution factor representing the registration accuracy requirement corresponding to each point cloud data point in the off-wall point structure set, the ICP registration process is further optimized by the model contribution factor; preferably, in some possible implementation methods of the embodiments of the present invention, the process of obtaining the registered building point cloud data includes: obtaining the minimization error function corresponding to the singular value decomposition method in the ICP point cloud registration; introducing the model contribution factor of each point cloud in each off-wall point cloud structure set as a weight into the minimization error function to determine the weighted minimization error function; continuing the ICP point cloud registration according to the weighted minimization error function to obtain the registered building point cloud data.

[0050] Among them, the minimization error function corresponding to the singular value decomposition method in ICP point cloud registration is a technical term well known to those skilled in the art, specifically: ;in, To minimize the error function; The first point in the source point cloud Point cloud data points; The target point cloud Matched nearest neighbors The rotation matrix between is the translation vector; is the number of point cloud data points in the source point cloud; further, the model contribution factor of each point cloud in each off-wall point cloud structure set is introduced as a weight into the minimization error function to determine the corresponding weighted minimization error function; specifically: ;in, is the weighted minimization error function; The first point in the source point cloud After determining the weighted minimization error function, the ICP algorithm is used to further perform point cloud registration to determine more accurate registration of building point cloud data. Finally, the BIM model of the corresponding building is constructed based on the registered building point cloud data.

[0051] In summary, the BIM building model construction method for building construction is first based on the plane continuity characteristics of the main wall structure in the building construction environment, and the point cloud is roughly segmented through the RANSAC multi-plane fitting algorithm to preliminarily separate the wall area and the external points of the complex structure; further for complex structure areas such as curved surfaces and arc connection areas, by calculating the relative distance and geometric coherence between the external points and the fitting plane, the model contribution factor is dynamically generated, and the point cloud in the high curvature area is given a higher weight; combined with the weighted ICP algorithm to optimize the alignment process, the key structural areas are aligned first, thereby reducing the damage to the integrity and continuity of the complex structure, making the registered building point cloud data more accurate after alignment, and improving the accuracy of the BIM model constructed based on the registered building point cloud data.

[0052] This application also provides a BIM building model construction system for building construction, see Figure 2 , which shows a structural diagram of a BIM building model construction system for building construction provided by an embodiment of the present invention. The system includes: a data acquisition and preprocessing module 201, a wall external point fitting plane determination module 202 and a building point cloud data construction module 203.

[0053] The data acquisition and preprocessing module 201 is configured to obtain initial building point cloud data by scanning the building using at least two radars; determine an initial wall area in the initial building point cloud data based on a plane fitting method; wherein data points outside the initial wall area are defined as off-wall points; The wall-external point fitting plane determination module 202 is configured to determine all corresponding local wall edges based on the extension of similar normal vectors of wall points located at the edge of each initial wall region; determine each set of wall-external point structures based on the similarity of the distance distributions between each wall-external point and all local wall edges; and perform plane fitting based on the position of each wall-external point structure set relative to each local wall edge to determine the corresponding wall-external point fitting plane. The building point cloud data construction module 203 is used to determine the model contribution factor of each point cloud according to the complexity of the neighborhood structure of each point cloud in the wall external point structure set; and perform point cloud cleaning according to the model contribution factor to obtain registered building point cloud data.

[0054] It should be noted that the system provided in the above embodiment is merely an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the BIM building model construction system for construction and the BIM building model construction method for construction provided in the above embodiment are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0055] The present application also provides a computer device. Figure 3 , which shows a schematic diagram of the structure of a computer device provided by an embodiment of the present invention, the computer device includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302, wherein when the processor 302 executes the computer program 303, the computer device can execute any one of the BIM building model construction methods for building construction introduced above.

[0056] An embodiment of the present application also provides a computer program product. When the computer program product is run on a computer device, the computer device can execute any one of the BIM building model construction methods for building construction introduced above.

[0057] An embodiment of the present application also provides a computer-readable storage medium, which stores computer program code. When the computer program code runs on a computer device, the computer device can execute any of the BIM building model construction methods for building construction introduced above.

[0058] In the embodiments provided in the present application, it should be understood that the provided computer devices, computer program products and computer-readable storage media are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the methods provided above and will not be repeated here.

[0059] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0060] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A BIM building model construction method for building construction, characterized in that: The method comprises: Scanning the building with at least two radars to obtain initial building point cloud data; determining an initial wall area in the initial building point cloud data based on a plane fitting method; wherein data points outside the initial wall area are off-wall points; Based on the extension of similar normal vectors of the wall points located at the edge of each initial wall area, all corresponding local wall edges are determined; based on the similarity of the distance distribution between each external wall point and all local wall edges, each external wall point structure set is determined; based on the position of each external wall point structure set relative to each local wall edge, a plane fitting is performed to determine the corresponding external wall point fitting plane; According to the complexity of the neighborhood structure of each point cloud in the off-wall point structure set, a model contribution factor of each point cloud is determined; and point cloud cleaning is performed according to the model contribution factor to obtain registered building point cloud data.

2. A BIM building model construction method for building construction according to claim 1, characterized in that: The process of obtaining the local edge of the wall includes: Obtaining an initial wall edge of each initial wall area; screening out false edge points based on a standard deviation of normal vectors corresponding to wall points on the initial wall edge in the initial wall area; and obtaining a reference wall edge after screening out the false edge points from the initial wall edge. The invention relates to a method for determining a wall point in the reference wall edge as an edge determination point; connecting adjacent edge determination points in sequence in the order of the reference wall edge to determine a determination edge line representing the line between each edge determination point and the next edge determination point; using the angle between the determination edge line of each edge determination point and the determination edge line of the previous edge determination point as the determination angle of each edge determination point; removing the determination edge line of each edge determination point when the determination angle of the edge determination point is greater than a preset first determination threshold, or when the difference between the determination angle of the corresponding edge determination point and the determination angle of the previous edge determination point is greater than a preset second determination threshold; and dividing the reference wall edge into at least two local wall edges after traversing all edge determination points.

3. A BIM building model construction method for building construction according to claim 2, characterized in that: The process of obtaining the false edge points includes: In each initial wall area, the corresponding reference normal vector is determined based on the mean of the normal vectors of all other wall points outside the initial wall edge; the angle between the normal vector of each wall point in the initial wall edge and the reference normal vector is used as the corresponding edge judgment angle; and the wall point whose corresponding edge judgment angle is less than a preset angle threshold is regarded as a false edge point.

4. A BIM building model construction method for building construction according to claim 1, characterized in that: The process of obtaining the wall external point structure set includes: A clustering space is constructed with the minimum Euclidean distance between an off-wall point and each local edge of the wall as each dimension; a clustering coordinate point of each off-wall point in the clustering space is determined according to the minimum Euclidean distance between each off-wall point and each local edge of the wall; a k-means clustering analysis is performed on all clustering coordinate points to obtain at least two clusters; and a set corresponding to all off-wall points corresponding to each clustering coordinate point in each cluster is used as a set of off-wall point structures.

5. The method for constructing a BIM building model for building construction according to claim 1, wherein: The process of obtaining the wall outer point fitting plane includes: Determine the reference relative distance of each local edge of the wall surface based on the average distance between all external wall points in each external wall point structure set and each local edge of the wall surface; determine the boundary position distance based on the overall change trend of the reference relative distance corresponding to each external wall point structure set; and arrange all reference relative distances of each external wall point structure set in ascending order to obtain a reference relative distance sequence; When the number of reference relative distances before the demarcation position distance is less than a preset number threshold, the local wall edges corresponding to the demarcation position distance in the reference relative distance sequence and all the local wall edges corresponding to all the reference relative distances before the demarcation position distance are used as the fitted wall edges of the corresponding external wall point structure set; when the number of reference relative distances before the demarcation position distance is greater than or equal to the preset number threshold, all the local wall edges corresponding to all the reference relative distances before the demarcation position distance in the reference relative distance sequence are used as the fitted wall edges of the corresponding external wall point structure set; According to each wall external point structure set and all corresponding fitted wall edges, plane fitting is performed using the RANSAC point cloud multi-plane fitting segmentation algorithm to determine the wall external point fitting plane corresponding to each wall external point structure set.

6. A BIM building model construction method for building construction according to claim 5, characterized in that: The process of obtaining the boundary position distance includes: In the reference relative distance sequence, the difference between each reference relative distance and the previous reference relative distance is used as the distance span value of each reference relative distance; and the reference relative distance with the largest distance span value in the reference relative distance sequence is used as the boundary position distance.

7. The method for constructing a BIM building model for building construction according to claim 1, wherein: The process of obtaining the model contribution factor includes: In the wall point structure set of each wall point fitting plane, all point cloud data points are traversed by the KNN nearest neighbor algorithm to obtain respective fitting model sequences; each point cloud data point in each fitting model sequence is sequentially used as a target data point; the fitting model sequence in which the target data point is located is used as a target data point model; a model is constructed based on the target data point model to obtain a target fitting model; the target data point in the fitting model sequence in which the target data point is located is removed and the model is constructed to obtain a comparative fitting model; Determine the corresponding reference contribution degree based on the position distribution of each point cloud data point in the target point cloud sequence relative to the target fitting model and the wall point fitting plane; determine the model fitting degree of the target fitting model based on the average of the reference contribution degrees of all point cloud data points in the target point cloud sequence; Based on the principle of obtaining the model fitting degree of the target point cloud sequence, the model fitting degree of the comparison fitting model is calculated; the difference between the model fitting degree of the target fitting model and the model fitting degree of the comparison fitting model is normalized to determine the model contribution factor of the target data point.

8. A BIM building model construction method for building construction according to claim 7, characterized in that: The process of obtaining the reference contribution degree includes: In the target point cloud sequence, the minimum Euclidean distance from each point cloud data point to the target fitting model is used as the reference matching accuracy of each point cloud data point; the negative correlation mapping value of the minimum Euclidean distance from each point cloud data point to the wall fitting plane corresponding to the target point cloud sequence is used as the curvature complexity weight of each point cloud data point; and the corresponding reference contribution degree is determined according to the product between the reference matching accuracy and the curvature complexity weight.

9. The method for constructing a BIM building model for building construction according to claim 1, wherein: The acquisition process of the registered building point cloud data includes: Obtain a minimization error function corresponding to the singular value decomposition method in ICP point cloud registration; introduce the model contribution factor of each point cloud in each off-wall point cloud structure set as a weight into the minimization error function to determine a weighted minimization error function; continue ICP point cloud registration according to the weighted minimization error function to obtain registered building point cloud data.

10. A BIM building model construction system for building construction, characterized in that: The system comprises: A data acquisition and preprocessing module is configured to obtain initial building point cloud data by scanning the building using at least two radars; determine an initial wall area in the initial building point cloud data based on a plane fitting method; wherein data points outside the initial wall area are off-wall points; The module for determining the external wall point fitting plane is used to determine all corresponding local wall edges based on the extension of similar normal vectors of wall points located at the edge of each initial wall area; determine each external wall point structure set based on the similarity of the distance distribution between each external wall point and all local wall edges; and perform plane fitting based on the position of each external wall point structure set relative to each local wall edge to determine the corresponding external wall point fitting plane. The building point cloud data construction module is used to determine the model contribution factor of each point cloud according to the complexity of the neighborhood structure of each point cloud in the wall point structure set; and perform point cloud cleaning according to the model contribution factor to obtain aligned building point cloud data.

Images