Building modeling method, device and equipment and storage medium

Abstract

The invention provides a building modeling method and device, equipment and a storage medium, and relates to the technical field of modeling, and the method comprises the steps: obtaining a target point cloud corresponding to a target building; wherein the target point cloud comprises a plurality of target sampling points; sequentially determining a plurality of target fitting planes with the maximum number of corresponding inner points based on the target point cloud, and determining a plurality of plane feature groups of the plurality of target fitting planes; wherein the number of the inner points is the number of the inner points, the inner points are sampling points, the distance between the inner points and the plane is smaller than a preset distance threshold value, and the inner points corresponding to different target fitting planes are not repeated; and constructing a plurality of target plane models corresponding to the plurality of target fitting planes according to the plurality of plane feature groups, and constructing a target building model corresponding to the target building based on the plurality of target plane models. Therefore, the overall precision of the building model is improved, the building model can be subsequently adjusted according to the fine granularity of the plane, and the flexibility of adjusting the model is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of modeling technology, and in particular to a building modeling method, device, equipment and storage medium. Background Art

[0002] In the process of urban renewal, the renovation of old buildings provides a lower-cost and less-impact solution compared to demolition and reconstruction. However, the architectural drawings and data of old buildings are often incomplete, and the natural changes of the buildings themselves over the years make it difficult to design and rebuild them based solely on traditional drawings.

[0003] At present, the overall modeling of old buildings can be carried out based on the point cloud data of old buildings. However, the model obtained in this way corresponds to the entire building, and only the entire model can be adjusted. The flexibility of adjustment is poor and the accuracy of the model obtained is also low. Summary of the invention

[0004] In order to solve the above technical problems, the embodiments of the present disclosure provide a building modeling method, device, equipment and storage medium.

[0005] In a first aspect, the present disclosure provides a building modeling method, the method comprising: Obtaining a target point cloud corresponding to a target building; wherein the target point cloud includes a plurality of target sampling points; Based on the target point cloud, a plurality of target fitting planes corresponding to the largest number of inliers are sequentially determined, and a plurality of plane feature groups of the plurality of target fitting planes are determined; wherein the number of inliers is the number of inliers, the inliers are sampling points whose distances from the planes are less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated; A plurality of target plane models corresponding to the plurality of target fitting planes are constructed according to the plurality of plane feature groups, and a target building model corresponding to the target building is constructed based on the plurality of target plane models.

[0006] In a second aspect, the present disclosure provides a building modeling device, the device comprising: An acquisition module, used to acquire a target point cloud corresponding to a target building; wherein the target point cloud includes a plurality of target sampling points; A determination module, used to determine, based on the target point cloud, a plurality of target fitting planes corresponding to the largest number of inliers in turn, and determine a plurality of plane feature groups of the plurality of target fitting planes; wherein the number of inliers is the number of inliers, the inliers are sampling points whose distances from the planes are less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated; A construction module is used to construct a plurality of target plane models corresponding to the plurality of target fitting planes according to the plurality of plane feature groups, and to construct a target building model corresponding to the target building based on the plurality of target plane models.

[0007] In a third aspect, the present disclosure provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device implements the above method.

[0008] In a fourth aspect, the present disclosure provides a building modeling device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0009] In a fifth aspect, the present disclosure provides a computer program product, wherein the computer program product comprises a computer program / instructions, and the computer program / instructions implement the above method when executed by a processor.

[0010] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has at least the following advantages: The disclosed embodiment provides a building modeling method, which obtains a target point cloud corresponding to a target building; wherein the target point cloud includes a plurality of target sampling points; based on the target point cloud, a plurality of target fitting planes corresponding to the largest number of corresponding inliers are determined in sequence, and a plurality of plane feature groups of the plurality of target fitting planes are determined; wherein the number of inliers is the number of inliers, and the inliers are sampling points whose distance from the plane is less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated; a plurality of target plane models corresponding to the plurality of target fitting planes are constructed according to the plurality of plane feature groups, and a target building model corresponding to the target building is constructed based on the plurality of target plane models. In the above scheme, after obtaining a plurality of target point clouds corresponding to the target building, a plurality of target fitting planes are determined according to the number of inliers determined based on the target point cloud, so that the conversion of the overall granularity of the target building to the plane granularity is realized, and modeling is performed according to the platform feature group of the target fitting plane, so that a target building model constructed by the target plane model is obtained, so that the construction of a building model based on the plane as the basic unit is realized, and the plane is used as the basic unit for constructing the model, so that the overall accuracy of the building model is improved, and the building model can be adjusted with the finer granularity of the plane later, so as to improve the flexibility of adjusting the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0012] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0013] Figure 1 A schematic diagram of a flow chart of a building modeling method provided by an embodiment of the present disclosure; Figure 2 A schematic diagram of a flow chart of another building modeling method provided by an embodiment of the present disclosure; Figure 3 A schematic diagram of a target point cloud before removing outliers provided in an embodiment of the present disclosure; Figure 4 A schematic diagram of a target point cloud after removing outliers provided in an embodiment of the present disclosure; Figure 5 A schematic diagram of a flow chart of another building modeling method provided by an embodiment of the present disclosure; Figure 6 A schematic diagram of a target building model provided by an embodiment of the present disclosure; Figure 7 A schematic diagram of a target plane model provided by an embodiment of the present disclosure; Figure 8 A schematic diagram of another target plane model provided by an embodiment of the present disclosure; Fig. 9 A schematic diagram of the structure of a building modeling device provided in an embodiment of the present disclosure; Fig.10 A schematic diagram of the structure of a building modeling device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0014] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0015] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0016] With the significant increase in the permanent population and significant improvement in urban infrastructure, the number of old buildings has also increased, and the comprehensive management of old buildings has become increasingly important.

[0017] In the process of urban renewal, the renovation and renewal of old buildings provides a lower-cost and less-impact alternative to demolition and reconstruction. However, the architectural drawings and data of old buildings are often incomplete, and coupled with the natural changes of the buildings themselves over the years, it is difficult to design and rebuild them based solely on traditional drawings.

[0018] At present, the overall modeling of old buildings can be carried out based on the point cloud data of old buildings. However, the model obtained in this way corresponds to the entire building, and only the entire model can be adjusted. The flexibility of adjustment is poor and the accuracy of the model obtained is also low.

[0019] In order to solve at least one of the above technical problems, the building modeling method provided by the embodiment of the present disclosure is described below. In the embodiment of the present disclosure, the building modeling method can be executed by an electronic device. The electronic device can include a device with communication function such as a tablet computer, a desktop computer, a laptop computer, etc., and can also include a device simulated by a virtual machine or a simulator.

[0020] Figure 1 FIG. 1 is a flow chart of a building modeling method provided by an embodiment of the present disclosure. Figure 1 As shown, the building modeling method may include the following steps.

[0021] Step 101, obtaining a target point cloud corresponding to a target building; wherein the target point cloud includes a plurality of target sampling points.

[0022] The target building may be a building to be reverse constructed into a three-dimensional model. This embodiment does not limit the target building. For example, the target building may be a building with incomplete drawing data, or an old building built before a preset construction year. The target point cloud may be a point cloud based on which the target building model is constructed for the target building. The point cloud may be a collection of points that characterize the surface characteristics of the target building. The target sampling point may be a sampling point in the target point cloud. The sampling point may be a point that characterizes the surface characteristics of the target building.

[0023] In the disclosed embodiment, the building modeling method can be implemented based on Python, C++, and Point Cloud Library (PCL), and the reverse modeling of target buildings such as old buildings can be implemented based on the building modeling method. In this embodiment, the target point cloud corresponding to the target building can be predetermined, and the building modeling device can obtain the target point cloud; or the building modeling device can use Python to perform pre-processing such as down-sampling processing and filtering processing on the point cloud obtained by scanning the target building, and obtain a target point cloud suitable for surface reconstruction of the target building.

[0024] Figure 2A flow chart of another building modeling method provided by an embodiment of the present disclosure is shown as follows: Figure 2 As shown, in some embodiments of the present disclosure, the target point cloud may be obtained by downsampling the original point cloud. Specifically, obtaining the target point cloud corresponding to the target building includes: Step 201, obtaining an original point cloud obtained by scanning the target building, and determining a minimum bounding box of the original point cloud; wherein the original point cloud includes a plurality of original sampling points.

[0025] The original point cloud may be a point cloud obtained by scanning the target building in three dimensions using a scanner, etc. The original sampling points may be sampling points in the original point cloud. The minimum bounding box may be a cuboid with the smallest volume that completely contains the original point cloud.

[0026] In this embodiment, a scanner or the like can scan the target building to obtain an original point cloud corresponding to the target building. The building modeling device can obtain the original point cloud and construct a minimum bounding box corresponding to the original point cloud according to the maximum and minimum values ​​of the coordinates of the original point cloud in the X, Y, and Z directions.

[0027] Step 202: Divide the minimum bounding box into a plurality of grids according to a preset grid side length.

[0028] The grid side length may be the side length of the grid. The grid may be the smallest processing unit divided by the minimum bounding box, and the grid may be understood as a voxel. This embodiment does not limit the shape of the grid, for example, the grid shape may be a cube or a cuboid.

[0029] In this embodiment, the building modeling device can perform spatial segmentation on the minimum bounding box in the X, Y, and Z directions according to the preset grid side lengths to obtain multiple grids.

[0030] Step 203 , determining target sampling points in each grid according to the original sampling points in each grid, and grouping the target sampling points into a target point cloud.

[0031] Among them, the target sampling points can correspond to the grids one by one.

[0032] In the embodiment of the present disclosure, for each grid, the building modeling device can calculate the coordinates according to the original sampling points in the grid to determine a target sampling point in the grid. The set of multiple target sampling points corresponding to multiple grids is the target point cloud. Thus, the downsampling processing of the original point cloud is realized with the grid as the minimum processing unit, and the target sampling points that characterize the characteristics of the original sampling points in the grid are obtained in a one-to-one correspondence with the grid, while retaining the characteristics of the original point cloud, reducing the number of sampling points to be processed subsequently.

[0033] In some embodiments of the present disclosure, determining the target sampling points in each grid according to the original sampling points in each grid includes: The side length of the minimum bounding box is standardized in the longitudinal and vertical directions according to the grid side length to obtain the standard side length of the longitudinal axis and the standard side length of the vertical axis of the minimum bounding box; the index number corresponding to each original sampling point is determined according to the standard coordinates, the standard side length of the longitudinal axis and the standard side length of the vertical axis of each original sampling point; wherein the standard coordinates of the original sampling point are determined according to the coordinates of the original sampling point, the minimum vertex coordinates of the minimum bounding box, and the grid side length; the index range corresponding to each grid is determined according to the standard coordinates of the vertices corresponding to each grid, the standard side length of the longitudinal axis and the standard side length of the vertical axis; wherein the standard coordinates of the vertex are determined according to the coordinates of the vertex, the minimum vertex coordinates of the minimum bounding box, and the grid side length; among the original sampling points, the grid sampling points in each grid are determined according to the index range and the index number; for each grid, the coordinates of the target sampling point are calculated according to the coordinates of the grid sampling points in the grid.

[0034] The standard side length of the longitudinal axis may be the side length obtained by normalizing the side length of the minimum bounding box according to the grid side length in the longitudinal dimension, and the standard side length of the longitudinal axis may be understood as the number of grids in the longitudinal direction. The standard side length of the vertical axis may be the side length obtained by normalizing the side length of the minimum bounding box according to the grid side length in the vertical dimension, and the standard side length of the vertical axis may be understood as the number of grids in the vertical direction. The standardization process may be a length conversion process using the corresponding grid side length as the basic unit.

[0035] The standard coordinates may be coordinates obtained by normalizing the original coordinates according to the minimum vertex coordinates of the minimum bounding box and the grid side length. The minimum vertex coordinates may be coordinates where the coordinate values ​​of each dimension in the minimum bounding box are all minimum values. The index number may be used to uniquely identify the original sampling point, and the index number may also be used to characterize the position of the original sampling point. The index range may correspond to the grid one-to-one, and the index range may be the range of index numbers of the original sampling points in the corresponding grid. The grid sampling point may be the original sampling point located inside the grid.

[0036] In this embodiment, the building modeling device can calculate the difference between the maximum and minimum values ​​of the original sampling point cloud in the three directions of X, Y, and Z to obtain the side lengths of the minimum bounding box in the three directions of X, Y, and Z. Taking the grid as a cube as an example, further, the modeling device can obtain the grid side length r of the grid, divide the corresponding side lengths of the minimum bounding box by the grid side length r in the three directions of X, Y, and Z, and round down to obtain the standard side length of the horizontal axis, the standard side length of the vertical axis, and the standard side length of the vertical axis. Furthermore, the coordinates can be standardized according to the minimum vertex coordinates of the minimum bounding box and the grid side length to obtain standard coordinates. The formula for standardizing the coordinates can be as follows: ; in, Indicates the minimum value of the horizontal coordinate of the minimum bounding box; Indicates the minimum ordinate value of the minimum bounding box; Indicates the minimum vertical coordinate value of the minimum bounding box; Represents the grid side length. For the coordinates of the original sampling point, represents the horizontal coordinate of the coordinate; represents the ordinate of the coordinate; Indicates the vertical coordinate of the coordinate; The horizontal coordinate of the coordinate in the standard coordinate system; The ordinate in the standard coordinates representing the coordinate; Indicates the vertical coordinate of the coordinate in the standard coordinate. For the coordinates of the vertex, represents the horizontal coordinate of the coordinate; represents the ordinate of the coordinate; Indicates the vertical coordinate of the coordinate; The horizontal coordinate of the standard coordinate system of the coordinate system; The ordinate in the standard coordinates representing the coordinate; Represents the vertical coordinate of the coordinate in standard coordinates.

[0037] For each original sampling point, the index number of the original sampling point Can be: ; in, Represents the abscissa of the original sampling point in standard coordinates; Represents the ordinate of the original sampling point in standard coordinates; Represents the vertical coordinate of the original sampling point in standard coordinates, represents the standard side length of the vertical axis, Indicates the standard side length of the vertical axis.

[0038] In this embodiment, after determining the standard coordinates of the vertex, the building modeling device can convert the standard coordinates of the vertex into the index number of the vertex according to the standard coordinates of the vertex, the standard side length of the longitudinal axis, and the standard side length of the vertical axis, in the same manner as converting the standard coordinates of the original sampling point into the index number of the original sampling point, and calculate the index range corresponding to the grid according to the index numbers of the 8 vertices in the grid.

[0039] Furthermore, the building modeling device can sort the index numbers corresponding to the original sampling points from small to large. For each grid, the original sampling points with index numbers within the index range corresponding to the grid are determined as the grid sampling points corresponding to the grid, and the centroid of the grid sampling points is calculated, and the centroid is used as the target sampling point corresponding to the grid.

[0040] In the above scheme, the original sampling points in the grid are efficiently determined by determining the index number and the index range. The original sampling points are downsampled based on voxels to eliminate redundant information and noise points. The accuracy of the final modeling is guaranteed while retaining the geometric features of the point cloud. The number of sampling points that need to be processed is reduced, the modeling efficiency is improved, and the problems of long calculation time and overfitting are solved. For example, in one embodiment, the number of original sampling points is nearly 20 million. After downsampling through a grid with a side length of 0.05, the number of target sampling points is reduced to more than one million, which significantly reduces the number of sampling points.

[0041] In an optional implementation, in a Python environment, the building modeling device can import Open3D and NumPy libraries. And create a point cloud reading function, which reads a txt file containing the original point cloud through the loadtxt method in the NumPy library, and automatically processes the delimiters of the coordinates in the txt file. After reading the txt file, the data stored in the file is converted into a target point cloud processed by the Open3D library, and the target point cloud is saved as a ply file through the point cloud writing function. Among them, the path of the txt file and the path of the ply file can be pre-configured, and the file format can be changed and the file can be saved based on the point cloud reading function.

[0042] After obtaining the ply file, the target point cloud stored in the file is subjected to voxel downsampling. Open3D and NumPy libraries are introduced again, and the building modeling device can read the ply file through the point cloud reading function. The target point cloud is downsampled through the voxel downsampling function, and the voxel size can be set to 0.05 meters. Voxel downsampling can greatly reduce the number of sampling points for subsequent processing and improve the efficiency of subsequent processing. After completing the downsampling, the sampling points obtained by the downsampling are saved in pcd format through the point cloud writing function.

[0043] When measuring the target building by 3D laser scanning, measurement noise is inevitable. The error caused by the measurement noise will produce sparse outliers, which will cause complex calculations and produce erroneous values ​​when determining the local features of the point cloud (for example, determining the normal vector), thereby causing the failure of surface modeling.

[0044] In some embodiments of the present disclosure, after downsampling the original point cloud, in order to filter out outliers in the target point cloud, the downsampled target point cloud may be filtered. In an optional implementation, after obtaining the target point cloud, the building modeling method further includes: Each target sampling point is determined as a sampling point to be processed, and a grid whose distance from the grid to be processed where the sampling point to be processed is less than a preset grid distance is determined as an associated grid; the number of target sampling points in the associated grid is determined to obtain an associated number, and if the associated number is less than a first number threshold, the sampling point to be processed is deleted; if the associated number is not less than the first number threshold, the sampling point to be processed is retained, or the target sampling point in the key grid whose distance from the sampling point to be processed is less than the preset sampling point distance is determined as an associated sampling point, and if the number of the associated sampling points is less than a second number threshold, the sampling point to be processed is deleted; if the number of the associated sampling points is not less than the second number threshold, the sampling point to be processed is retained.

[0045] In another optional implementation, since the outliers in the point cloud are usually sparsely distributed in space, the building modeling device can identify the outliers by identifying the low-density part in the target point cloud, pre-set the density threshold, and delete the target sampling points with density values ​​lower than the density threshold. Specifically, for each target sampling point, a statistical analysis is performed on the neighborhood of the target sampling point, and the average distance between each target sampling point and its nearest k points is calculated. The average distance characterizes the density of the neighborhood point cloud corresponding to the target sampling point, and k can be a set positive integer. Assume that the average distance obeys a Gaussian distribution, which is described by a mean μ and a standard deviation σ. The standard deviation multiple std and a positive integer k are pre-set. If the average distance from the target sampling point to its k neighboring target sampling points is within the interval (μ-σ·std, μ+σ·std), the target sampling point is retained; otherwise, it is indicated that the target sampling point is an outlier and the target sampling point is deleted.

[0046] In an optional implementation, the Open3D library is pre-imported in the Python environment, and the model building device can load the file recording the target point cloud through the point cloud reading function, and assign the target point cloud to the point cloud variable pcd. The number of sampling points in the current target point cloud can be viewed by printing the point cloud variable pcd. Subsequently, the statistical filtering processing is implemented through the discrete point removal function, and the standard deviation multiple and the number of closest points need to be set for the discrete point removal function. The number of closest points defines the number of nearest points considered when calculating the average distance corresponding to each target sampling point, and the number of closest points can be set to 150. The standard deviation multiple sets the standard deviation threshold for determining whether the target sampling point is an outlier, and the standard deviation multiple can be set to 2. Furthermore, by comparing the average distance between the target sampling point and other target sampling points in its neighborhood, outliers are identified and removed according to the standard deviation of the Gaussian distribution.

[0047] After completing the point cloud filtering, the building modeling device can assign the point cloud after filtering outliers to a new point cloud variable, save the new point cloud variable in pcd format through the point cloud write function, and use the new point cloud variable as the current target point cloud. The pcd format data can be used to observe the target point cloud through the point cloud comparison (CloudCompare) tool. Figure 3 A schematic diagram of a target point cloud before removing outliers provided in an embodiment of the present disclosure, Figure 4 A schematic diagram of a target point cloud after removing outliers provided in an embodiment of the present disclosure. Figure 3 and Figure 4 By comparison, it can be determined that the outliers have been effectively removed. By removing the outliers, the efficiency of subsequent work such as surface reconstruction can be effectively improved.

[0048] In the above scheme, the sampling points are filtered to automatically remove outliers in the point cloud, retain valid information in the point cloud, improve the data quality and accuracy of the retained point cloud, and create a foundation for the subsequent accurate establishment of the target plane model.

[0049] Step 102, based on the target point cloud, determine in turn multiple target fitting planes corresponding to the largest number of inliers, and determine multiple plane feature groups of the multiple target fitting planes; wherein the number of inliers is the number of inliers, and the inliers are sampling points whose distances to the planes are less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated.

[0050] The number of inliers corresponding to the target fitting plane may be the number of target sampling points whose distance from the target fitting plane is less than a preset distance threshold. The preset distance threshold may be a pre-set distance threshold for screening inliers, and this embodiment does not limit the preset distance threshold. The target fitting plane may be a fitting plane determined based on the positional relationship between the plane and the target sampling points in the target point cloud.

[0051] A plane feature group may be a set of feature values ​​that characterize the target fitting plane features. This embodiment does not impose any restrictions on the plane feature group. In some embodiments of the present disclosure, the plane feature group includes a plane boundary, a plane normal vector, and a reference point of the target fitting plane. The plane boundary may be used to determine the outline of the target fitting plane, and the plane normal vector may be used to determine the direction of the target fitting plane. The plane normal vector may be used to characterize the possible tilt problem of the target building. The reference point may be a point through which the target fitting plane passes. This embodiment does not impose any restrictions on the reference point. For example, the reference point may be the intersection of the target fitting plane and the plane normal vector.

[0052] In the disclosed embodiment, the building modeling device can perform plane-based point cloud segmentation and plane feature group extraction on the target point cloud through C++ and the point cloud data processing library. Specifically, since the surface of the target building can be approximately considered as a plane, in this embodiment, the building modeling device can randomly select multiple target sampling points in the target point cloud to fit the fitting plane, and count the number of inner points in the target point cloud that are the inner points of the fitting plane, determine the fitting plane with the largest number of corresponding inner points as the target fitting plane, delete the inner points corresponding to the target fitting plane, return to re-select the target sampling points, and iterate the cycle to determine the target fitting plane until the preset end condition is met.

[0053] After determining the target fitting plane, the building modeling device can determine the plane normal vector and reference point of the target fitting plane, and calculate the projection point of the interior point on the target fitting plane through the boundary algorithm to determine the plane boundary of the target fitting plane, wherein the boundary algorithm can be used to determine the algorithm of the boundary formed by the sampling points, and this embodiment does not limit the boundary algorithm, for example, the boundary algorithm can be a convex hull algorithm. And the plane normal vector, reference point, and plane boundary are used as the plane feature group of the target fitting plane.

[0054] In some embodiments of the present disclosure, the plane normal vector may be a normal vector determined based on the inner points of the target fitting plane. Determining the plane normal vector through the inner points can avoid deviations caused by local points in the process of modeling the plane, thereby improving the fitting accuracy of the target plane model. The reference point may be a point determined based on the inner points of the target fitting plane. Specifically, the building construction device may determine the plane normal vector using the least-binary method for the inner points corresponding to the target fitting plane. In the least-binary method, the building modeling device may construct a covariance matrix based on the inner points corresponding to the target fitting plane, perform eigenvalue decomposition on the covariance matrix, and obtain the plane normal vector of the target fitting plane. Furthermore, the coordinate average of the inner points corresponding to the target fitting plane is calculated to obtain the inner point centroid, and the inner point centroid is used as the reference point of the target fitting plane.

[0055] In an optional implementation, the process of determining the plane normal vector and the reference point includes: performing eigenvalue decomposition of the covariance matrix corresponding to the interior point through an eigenvalue decomposition function. In this process, the eigenvector of the covariance matrix represents the direction of the principal axis, and the eigenvector corresponding to the smallest eigenvalue represents the plane normal vector. If the direction of the calculated plane normal vector does not point to the positive Z axis, the plane normal vector needs to be reversely processed to ensure its correctness.

[0056] Figure 5 A flowchart of another building modeling method provided by the embodiment of the present disclosure is shown as follows: Figure 5 As shown, in some embodiments of the present disclosure, multiple target fitting planes corresponding to the largest number of inner points are sequentially determined based on the target point cloud, including: Step 501: determine the target point cloud as an intermediate point cloud; wherein the intermediate point cloud includes a plurality of intermediate sampling points.

[0057] The intermediate point cloud may be a point cloud determined in an intermediate process of determining the target fitting plane based on the target point cloud, and the intermediate sampling point may be a sampling point in the intermediate point cloud.

[0058] In this embodiment, the building modeling apparatus may determine the target point cloud as an intermediate point cloud, and determine the target acquisition point in the target point cloud as an intermediate acquisition point.

[0059] Step 502: Select multiple groups of fitting sampling points from the middle sampling points, and obtain multiple fitting planes by fitting the multiple groups of fitting sampling points.

[0060] Among them, a group of fitting sampling points can be composed of multiple intermediate sampling points. The number of intermediate sampling points in a group of fitting sampling points is not limited in this embodiment, for example, it can be 3. Each group of fitting sampling points corresponds to the fitting plane one by one. This embodiment does not limit the number of groups of the fitting sampling points, and the number of groups can be set according to user needs, etc. By setting an appropriate upper limit on the number of groups, the accuracy of the target fitting plane can be improved while reducing unnecessary calculations and improving the speed of determining the plane. By setting an appropriate number of groups, the target fitting plane that meets the actual situation can be obtained while reducing the amount of calculation.

[0061] In this embodiment, a preset number of intermediate sampling points are randomly selected from the intermediate sampling points, and the preset number of intermediate sampling points are used as a group of fitting sampling points. Plane fitting is performed for each group of fitting sampling points to obtain a fitting plane corresponding to the group of fitting sampling points.

[0062] Step 503: Determine the number of multiple interior points corresponding to the multiple fitting planes.

[0063] In this embodiment, multiple groups of fitting sampling points correspond to multiple fitting planes. For each fitting plane, the middle sampling points whose distance to the fitting plane is less than a preset distance threshold are determined as the inner points corresponding to the fitting plane, and the number of the inner points is counted to obtain the number of inner points corresponding to the fitting plane.

[0064] Step 504: determine the fitting plane corresponding to the maximum number of inliers as the target fitting plane.

[0065] In this embodiment, after determining the numbers of multiple inliers corresponding to the multiple fitting planes, the fitting plane corresponding to the maximum number of the multiple inliers is determined as the target fitting plane corresponding to the current intermediate point cloud.

[0066] In the disclosed embodiment, the building modeling device can randomly select multiple intermediate sampling points in the intermediate point cloud as a group of fitting sampling points, and fit the corresponding fitting plane to the group of fitting sampling points. For the fitting plane, determine the intermediate sampling points whose distance from the fitting plane is less than a preset distance threshold, use the intermediate sampling points as the inner points of the fitting plane, and count the number of inner points of the inner points. The building modeling device determines multiple groups of fitting sampling points. In each iterative determination process, if the number of inner points corresponding to the current fitting plane is greater than that of the previous fitting plane, the current fitting plane is determined as the current optimal plane. This process is repeated until the number of iterations reaches the preset number of groups of fitting sampling points, and the optimal plane at this time is determined as the target fitting plane corresponding to the intermediate point cloud.

[0067] In the above scheme, the plane segmentation of the intermediate point cloud is implemented based on the random sampling consistency algorithm, which can effectively distinguish noise and outliers and has good robustness to noise and outliers. By randomly determining the fitting sampling points and performing iterative optimization, the best target fitting plane with the most internal points can be extracted from the intermediate point cloud. While processing the intermediate point cloud well, it can converge faster and improve the accuracy of the plane results.

[0068] Step 505, delete the inner points corresponding to the target fitting plane in the intermediate point cloud to obtain a new intermediate point cloud, and return to determine the target fitting plane corresponding to the new intermediate point cloud until a preset end condition is met.

[0069] The preset end condition may be a preset condition for stopping the iteration to determine the target fitting plane, and this embodiment does not limit the preset end condition. For example, the preset end condition may be that the target fitting plane reaches a preset number of planes, or the preset end condition may be that the number of sampling points in the new intermediate point cloud is less than a preset sampling point number threshold.

[0070] In this embodiment, after determining the target fitting plane corresponding to the current intermediate point cloud, the inner points of the target fitting plane are deleted in the current intermediate point cloud to obtain a new intermediate point cloud, and based on the new intermediate point cloud, the process returns to step 502 to determine the target fitting plane corresponding to the new intermediate point cloud until the preset end condition is met.

[0071] In the above scheme, the target fitting plane is first determined by random sampling, and then the plane normal vector and other features of the target fitting plane are determined by the least binary method, which not only improves the accuracy of simulation modeling of the exterior wall of the target building, but also reduces the difficulty and cost of modeling.

[0072] Step 103: construct multiple target plane models corresponding to the multiple target fitting planes according to the multiple plane feature groups, and construct a target building model corresponding to the target building based on the multiple target plane models.

[0073] The target plane model may be a three-dimensional virtual model corresponding to the target fitting plane. The target building model may be a three-dimensional virtual model of the target building, and the smallest component unit of the target building model may be a target plane model. The target building model may be a building information model (Building Information Modeling, BIM) of the target building.

[0074] In the disclosed embodiment, after determining a plurality of plane feature groups, a corresponding target plane model may be constructed automatically or manually according to the plane feature groups, and a target building model formed by splicing the target plane models may be determined. Figure 6A schematic diagram of a target building model provided by an embodiment of the present disclosure. Figure 6 As shown, multiple target plane models are spliced ​​together to generate the target building model.

[0075] In some embodiments of the present disclosure, multiple target plane models corresponding to multiple target fitting planes are constructed based on multiple plane feature groups, including: displaying the plane feature group corresponding to each target fitting plane on a preset interface; and constructing the target plane model in response to the plane construction operation performed by the user based on the plane feature group.

[0076] The preset interface may be a pre-set display interface. The plane construction operation may be an operation in which a user constructs a target plane model according to a plane feature group through a program.

[0077] In this embodiment, after determining the plane feature group, the architectural modeling device can display the plane feature group to the user in a visual manner on a preset interface. After seeing the plane feature group, the user performs a plane construction operation according to the plane feature group, and the architectural modeling device constructs a corresponding target plane in response to the plane construction operation. Figure 7 A schematic diagram of a target plane model provided in an embodiment of the present disclosure, Figure 8 FIG. 1 is a schematic diagram of another target plane model provided by an embodiment of the present disclosure. Figure 7 As shown in , the target plane model corresponds to the front of the target building, such as Figure 8 As shown, the target plane model corresponds to the back side of the target building.

[0078] In the above scheme, a plane feature group is determined by segmenting the point cloud by a plane, and the plane is reconstructed manually based on the plane feature group. Compared with the modeling method of directly modeling the model corresponding to the target building as a whole based on the point cloud in the related technology, the target building is modeled with a finer granularity. The final target building model has higher accuracy and avoids directly constructing the plane based on a large number of point clouds, thereby improving the construction efficiency of the target building model.

[0079] In the building modeling method provided by the embodiment of the present disclosure, a target point cloud corresponding to a target building is obtained; wherein the target point cloud includes a plurality of target sampling points; based on the target point cloud, a plurality of target fitting planes corresponding to the largest number of corresponding inliers are determined in sequence, and a plurality of plane feature groups of the plurality of target fitting planes are determined; wherein the number of inliers is the number of inliers, and the inliers are sampling points whose distance from the plane is less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated; a plurality of target plane models corresponding to the plurality of target fitting planes are constructed according to the plurality of plane feature groups, and a target building model corresponding to the target building is constructed based on the plurality of target plane models. In the above scheme, after obtaining a plurality of target point clouds corresponding to the target building, a plurality of target fitting planes are determined according to the number of inliers determined based on the target point cloud, so that the conversion of the overall granularity of the target building to the plane granularity is realized, and modeling is performed according to the platform feature group of the target fitting plane, so that a target building model constructed by the target plane model is obtained, so that the construction of a building model based on a plane as a basic unit is realized, and the plane is used as the basic unit for constructing the model, so that the overall accuracy of the building model is improved, and the building model can be adjusted with the finer granularity of the plane later, so as to improve the flexibility of adjusting the model.

[0080] In some embodiments of the present disclosure, the building modeling method further includes: performing surface reconstruction processing on the target point cloud by using a surface reconstruction algorithm to obtain a reference building model; and displaying the reference building model for comparison next to the construction area of ​​the target building model.

[0081] The surface reconstruction algorithm may be an algorithm for reconstructing the target building as a whole as the minimum basic unit based on the target point cloud. This embodiment does not limit the surface reconstruction algorithm. For example, the surface reconstruction algorithm may be a ball rotation algorithm. The reference model building model may be a three-dimensional virtual model used as a reference in the process of the user constructing the target building model. The construction area may be an area for constructing the target building model.

[0082] In this embodiment, the building modeling device can perform surface reconstruction processing on the target point cloud through Python to obtain a reference building model generated by reverse modeling of the target building. Specifically, the building modeling device can import the Open3D library, read the target point cloud through the point cloud reading function and assign the target point cloud to the point cloud variable pcd, and print the point cloud variable pcd to output the number of points in the target point cloud. The normal line is calculated for the target point cloud through the normal vector estimation function, and the adjacent points are found through a hybrid search of the k-dimensional tree (kd-tree). The search radius is set to 0.5 meters, and only 300 sampling points in the neighborhood are considered. After the normal line calculation is completed, a radius list is defined. For example, the radius list can be [0.5, 0.1, 0.2, 0.5]. The numbers in the radius list represent the radius of the three-dimensional sphere used for surface reconstruction. The unit is consistent with the target building model, which can be meters. Then, a ball rotation algorithm based on point cloud is used to create a triangular mesh function, and a triangular mesh is created using the ball rotation algorithm to achieve surface reconstruction to obtain a reference building model. The model writing function is used to save the reference building model in an obj format that can be imported into Revit.

[0083] Furthermore, the building modeling device can read the reference building model and display the reference building model next to the construction area of ​​the target building model. In the process of building the target building model, the user can compare the reference building model with the unfinished target building model, so that the user can understand whether there are errors in his plane construction operation, etc., which lead to errors in the construction of the target building model.

[0084] In some embodiments of the present disclosure, the building modeling method includes: preprocessing the original point cloud through Python to obtain a target point cloud, and the preprocessing includes downsampling processing and filtering processing. Through the preprocessing, redundant sampling points in the target point cloud can be eliminated. The target point cloud is segmented through C++ and the point cloud data processing library to obtain a target fitting plane, and the plane boundary corresponding to the target fitting plane is determined. The inner point and plane boundary corresponding to the target fitting plane are exported through C++. The inner point is calculated by the least binary method through Python to determine the plane normal vector corresponding to the inner point, and the average coordinate of the inner point is calculated to obtain the corresponding reference point. The surface is reconstructed according to the plane normal vector, the reference point and the plane boundary to obtain the target plane model, and the target plane model can be imported into the Revit software. In the Revit software, the reverse modeling of the target building is realized based on the target plane model to generate the target building model. The architectural modeling method provided by the embodiment of the present disclosure realizes the automation of the main process of constructing the architectural model, and determines the plane feature group corresponding to the plane, and realizes the reverse modeling of the target architectural model based on the plane feature group. Compared with the reverse modeling method of directly remodeling based on the point cloud, the modeling error obtained by the sampling point error is reduced, and the modeling efficiency is improved.

[0085] Based on the above method embodiment, the present disclosure also provides a building modeling device, referring to Fig. 9 , is a schematic diagram of the structure of a building modeling device provided by an embodiment of the present disclosure, the device comprising: The acquisition module 901 is used to acquire a target point cloud corresponding to a target building; wherein the target point cloud includes a plurality of target sampling points; A determination module 902 is used to sequentially determine a plurality of target fitting planes corresponding to the largest number of inliers based on the target point cloud, and determine a plurality of plane feature groups of the plurality of target fitting planes; wherein the number of inliers is the number of inliers, the inliers are sampling points whose distances from the planes are less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated; The construction module 903 is used to construct a plurality of target plane models corresponding to the plurality of target fitting planes according to the plurality of plane feature groups, and to construct a target building model corresponding to the target building based on the plurality of target plane models.

[0086] In an optional implementation, the acquisition module 901 includes: An acquisition submodule, used to acquire an original point cloud obtained by scanning the target building, and determine a minimum bounding box of the original point cloud; wherein the original point cloud includes a plurality of original sampling points; A division submodule, used for dividing the minimum bounding box into a plurality of grids according to a preset grid side length; The determination submodule is used to determine the target sampling points in each of the grids according to the original sampling points in each of the grids, and to aggregate the target sampling points into the target point cloud.

[0087] In an optional implementation manner, determining the target sampling points in each of the grids according to the original sampling points in each of the grids includes: Standardizing the side length of the minimum bounding box in the longitudinal direction and the vertical direction according to the side length of the grid to obtain the longitudinal axis standard side length and the vertical axis standard side length of the minimum bounding box; Determine the index number corresponding to each of the original sampling points according to the standard coordinates of each of the original sampling points, the standard side length of the longitudinal axis, and the standard side length of the vertical axis; wherein the standard coordinates of the original sampling points are determined according to the coordinates of the original sampling points, the minimum vertex coordinates of the minimum bounding box, and the side length of the grid; Determine the index range corresponding to each grid according to the standard coordinates of the vertices corresponding to each grid, the standard side length of the longitudinal axis, and the standard side length of the vertical axis; wherein the standard coordinates of the vertex are determined according to the coordinates of the vertex, the minimum vertex coordinates of the minimum bounding box, and the side length of the grid; In the original sampling points, according to the index range and the index number, a grid sampling point in each grid is determined; For each of the grids, the coordinates of the target sampling point are calculated according to the coordinates of the grid sampling points in the grid.

[0088] In an optional implementation manner, the determining, based on the target point cloud, a plurality of target fitting planes corresponding to the largest number of inner points in sequence includes: Determine the target point cloud as an intermediate point cloud; wherein the intermediate point cloud includes a plurality of intermediate sampling points; Selecting multiple groups of fitting sampling points from the intermediate sampling points, and obtaining multiple fitting planes by fitting the multiple groups of fitting sampling points; Determine the number of multiple interior points corresponding to multiple fitting planes; Determine the fitting plane corresponding to the maximum value of the number of inliers as the target fitting plane; The inner points corresponding to the target fitting plane in the intermediate point cloud are deleted to obtain a new intermediate point cloud, and the target fitting plane corresponding to the new intermediate point cloud is determined again until a preset end condition is met.

[0089] In an optional implementation, the plane feature group includes a plane boundary, a plane normal vector, and a reference point of the target fitting plane.

[0090] In an optional implementation manner, constructing multiple target plane models corresponding to the multiple target fitting planes according to the multiple plane feature groups includes: Displaying the plane feature group corresponding to each target fitting plane on a preset interface; In response to a plane construction operation performed by a user based on the plane feature group, the target plane model is constructed.

[0091] In an optional implementation manner, the building modeling device further includes: A reconstruction module, used for performing surface reconstruction processing on the target point cloud by using a surface reconstruction algorithm to obtain a reference building model; The display module is used to display the reference building model next to the construction area of ​​the target building model for comparison.

[0092] It should be noted that Fig. 9 The illustrated building modeling device can execute each step in the above-mentioned building modeling method embodiment, and realize each process and effect in the above-mentioned building modeling method embodiment, which will not be described in detail here.

[0093] In addition to the above-mentioned method and apparatus, the embodiments of the present disclosure further provide a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal device, the terminal device implements the building modeling method described in the embodiments of the present disclosure.

[0094] The embodiment of the present disclosure further provides a computer program product, which includes a computer program / instructions. When the computer program / instructions are executed by a processor, the building modeling method described in the embodiment of the present disclosure is implemented.

[0095] In addition, the present disclosure also provides a building modeling device, see Fig.10 As shown, it may include: Processor 1001, memory 1002, input device 1003 and output device 1004. The number of processors 1001 in the building modeling device can be one or more. Fig.10 In some embodiments of the present disclosure, the processor 1001, the memory 1002, the input device 1003 and the output device 1004 may be connected via a bus or other means, wherein: Fig.10 The example of connecting through bus is taken in the following.

[0096] The memory 1002 can be used to store software programs and modules. The processor 1001 executes various functional applications and data processing of the building modeling device by running the software programs and modules stored in the memory 1002. The memory 1002 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function, etc. In addition, the memory 1002 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. The input device 1003 can be used to receive input digital or character information, and generate signal input related to user settings and function control of the building modeling device.

[0097] Specifically in this embodiment, the processor 1001 will load the executable files corresponding to the processes of one or more applications into the memory 1002 according to the following instructions, and the processor 1001 will run the applications stored in the memory 1002, so as to realize the various functions of the above-mentioned building modeling device.

[0098] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0099] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A building modeling method, characterized in that: The method comprises: Obtaining a target point cloud corresponding to a target building; wherein the target point cloud includes a plurality of target sampling points; Based on the target point cloud, a plurality of target fitting planes corresponding to the largest number of inliers are sequentially determined, and a plurality of plane feature groups of the plurality of target fitting planes are determined; wherein the number of inliers is the number of inliers, the inliers are sampling points whose distances from the planes are less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated; A plurality of target plane models corresponding to the plurality of target fitting planes are constructed according to the plurality of plane feature groups, and a target building model corresponding to the target building is constructed based on the plurality of target plane models.

2. The method according to claim 1, characterized in that The step of obtaining a target point cloud corresponding to the target building includes: Acquire an original point cloud obtained by scanning the target building, and determine a minimum bounding box of the original point cloud; wherein the original point cloud includes a plurality of original sampling points; Dividing the minimum bounding box into a plurality of grids according to a preset grid side length; The target sampling points in each of the grids are determined according to the original sampling points in each of the grids, and the target sampling points are aggregated into the target point cloud.

3. The method according to claim 2, characterized in that The step of determining the target sampling points in each of the grids according to the original sampling points in each of the grids comprises: Standardizing the side length of the minimum bounding box in the longitudinal direction and the vertical direction according to the side length of the grid to obtain the longitudinal axis standard side length and the vertical axis standard side length of the minimum bounding box; Determine the index number corresponding to each of the original sampling points according to the standard coordinates of each of the original sampling points, the standard side length of the longitudinal axis, and the standard side length of the vertical axis; wherein the standard coordinates of the original sampling points are determined according to the coordinates of the original sampling points, the minimum vertex coordinates of the minimum bounding box, and the side length of the grid; Determine the index range corresponding to each grid according to the standard coordinates of the vertices corresponding to each grid, the standard side length of the longitudinal axis, and the standard side length of the vertical axis; wherein the standard coordinates of the vertex are determined according to the coordinates of the vertex, the minimum vertex coordinates of the minimum bounding box, and the side length of the grid; In the original sampling points, according to the index range and the index number, a grid sampling point in each grid is determined; For each of the grids, the coordinates of the target sampling point are calculated according to the coordinates of the grid sampling points in the grid.

4. The method according to claim 1, characterized in that: The step of sequentially determining a plurality of target fitting planes corresponding to the largest number of inner points based on the target point cloud comprises: Determine the target point cloud as an intermediate point cloud; wherein the intermediate point cloud includes a plurality of intermediate sampling points; Selecting multiple groups of fitting sampling points from the intermediate sampling points, and obtaining multiple fitting planes by fitting the multiple groups of fitting sampling points; Determine the number of multiple interior points corresponding to multiple fitting planes; Determine the fitting plane corresponding to the maximum value of the number of inliers as the target fitting plane; The inner points corresponding to the target fitting plane in the intermediate point cloud are deleted to obtain a new intermediate point cloud, and the target fitting plane corresponding to the new intermediate point cloud is determined again until a preset end condition is met.

5. The method according to claim 1, characterized in that: The plane feature group includes a plane boundary, a plane normal vector, and a reference point of the target fitting plane.

6. The method according to claim 1, characterized in that The step of constructing a plurality of target plane models corresponding to the plurality of target fitting planes according to the plurality of plane feature groups includes: Displaying the plane feature group corresponding to each target fitting plane on a preset interface; In response to a plane construction operation performed by a user based on the plane feature group, the target plane model is constructed.

7. The method according to claim 6, characterized in that The method further comprises: Performing surface reconstruction processing on the target point cloud by using a surface reconstruction algorithm to obtain a reference building model; The reference building model is displayed for comparison next to the construction area of ​​the target building model.

8. A building modeling device, characterized in that: The device comprises: An acquisition module, used to acquire a target point cloud corresponding to a target building; wherein the target point cloud includes a plurality of target sampling points; A determination module, used to determine, based on the target point cloud, a plurality of target fitting planes corresponding to the largest number of inliers in turn, and determine a plurality of plane feature groups of the plurality of target fitting planes; wherein the number of inliers is the number of inliers, the inliers are sampling points whose distances from the planes are less than a preset distance threshold, and the inliers corresponding to different target fitting planes are not repeated; A construction module is used to construct a plurality of target plane models corresponding to the plurality of target fitting planes according to the plurality of plane feature groups, and to construct a target building model corresponding to the target building based on the plurality of target plane models.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device implements the method according to any one of claims 1 to 7.

10. A building modeling device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

Images