Building structure design three-dimensional model construction method based on spatial point cloud data

Abstract

The invention relates to the technical field of building model construction, in particular to a building structure design three-dimensional model construction method based on spatial point cloud data, and the method comprises the steps: dividing the spatial point cloud data of a construction building into a plurality of projection planes; dividing super-pixel clusters according to distribution characteristics and numerical characteristics of projection data points in a projection plane, and analyzing similarity among different super-pixel clusters to obtain clusters of the super-pixel clusters; then analyzing the position relationship between the superpixel clusters in the cluster and the volume difference condition and the volume distribution condition of the superpixel clusters in the cluster to obtain the noise influence degree of the projection plane; and finally, according to the noise influence degree, carrying out noise reduction processing on the spatial point cloud data. According to the method, different degrees of noise reduction processing is carried out on different projection planes and different clusters of spatial point cloud data of a construction building, so that the efficiency and accuracy of noise reduction processing are improved, and a reliable building structure design three-dimensional model can be constructed.

Description

Technical Field

[0001] The present invention relates to the technical field of building model construction, and in particular to a method for constructing a three-dimensional model of a building structure design based on spatial point cloud data. Background Art

[0002] Spatial point cloud data is three-dimensional spatial data collected and processed by laser scanning technology. It consists of a large number of discrete data points, each of which represents a spatial position. When spatial point cloud data is used for building three-dimensional models, it can accurately reconstruct the geometric shape of the building and provide detailed information for design, analysis and maintenance; when used in the construction process, it can monitor the progress of construction. Therefore, by collecting spatial point cloud data of the construction building and building an accurate three-dimensional model of the construction building structure, it can be used to accurately monitor the construction progress of the construction building.

[0003] In order to construct a three-dimensional model of the construction building, drones are generally used to collect spatial point cloud data of the construction building. However, due to the presence of many reflective objects in the construction environment, there is a lot of noise in the collected spatial point cloud data, so the collected spatial point cloud data needs to be denoised. When denoising the spatial point cloud data, due to the irregular distribution of the construction building structure, the noise impact on each part is different; in addition, due to the complex scene of the construction site, the complexity of the noise is increased, which in turn increases the difficulty of noise reduction.

[0004] Therefore, there is an urgent need for a method to accurately reduce the noise of the spatial point cloud data of the construction building in order to construct a reliable three-dimensional model for building structure design. Summary of the invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a method for constructing a three-dimensional model of building structure design based on spatial point cloud data.

[0006] According to a first aspect of an embodiment of the present invention, a method for constructing a three-dimensional model of a building structure design based on spatial point cloud data is provided, and the technical solution adopted is specifically as follows: Collect spatial point cloud data of buildings; Dividing the spatial point cloud data into data points of different grid surfaces, projecting the data points in each grid surface, and obtaining a projection plane corresponding to each grid surface; Analyzing the distribution characteristics and numerical characteristics of the projection data points in the projection plane, performing superpixel segmentation on the projection data points in the projection plane, and obtaining a plurality of superpixel clusters; Analyzing the similarities between the superpixel clusters based on the distribution characteristics and numerical characteristics of the projection data points inside the superpixel clusters to obtain clustering of the superpixel clusters; Analyzing the positional relationship between the superpixel clusters in the cluster to obtain a distribution factor of the cluster; Analyzing the volume difference and volume distribution of the superpixel clusters in the cluster to obtain the noise impact factor of the cluster; According to the distribution factor and the noise impact factor of the cluster, combined with the volume of the cluster, the noise impact degree of the projection plane is obtained; According to the noise impact degree, noise reduction processing is performed on the spatial point cloud data.

[0007] In some embodiments of the present invention, dividing the spatial point cloud data into data points of different grid surfaces includes: The spatial point cloud data is surface reconstructed using a triangular mesh model, and the mesh in the triangular mesh algorithm is divided into a plurality of mesh surfaces, and then the mesh surfaces are segmented using a region growing algorithm to divide the spatial point cloud data into data points on different mesh surfaces.

[0008] In some embodiments of the present invention, projecting the data points in each mesh surface to obtain the projection plane corresponding to each mesh surface includes: Project all data points in each grid surface onto a plane perpendicular to its viewing angle to obtain a projection plane corresponding to each grid surface, wherein the data value of the projected data point in the projection plane is the height value between the data point corresponding to the projection data point and the projection plane.

[0009] In some embodiments of the present invention, analyzing the distribution characteristics of the projection data points in the projection plane includes: In the projection plane, obtaining the first minimum distance between all projection data points and other projection data points, and recording the maximum value of all the first minimum distances as the first distance; For each projection data point in the projection plane, a circle is drawn in the projection plane with the projection data point as the center and the first distance as the radius to obtain the surrounding area of ​​the projection data point; The density of the projected data points in the surrounding area is analyzed to obtain distribution characteristics of the projected data points in the projection plane.

[0010] In some embodiments of the present invention, analyzing the numerical characteristics of the projection data points in the projection plane includes: In the surrounding area, obtaining the second minimum distances between all projection data points and other projection data points, and recording the maximum value of all the second minimum distances as the second distance; For any projected data point in the surrounding area, if its data value is the largest within the second distance range, the projected data point is recorded as a maximum value point, and all maximum value points in the surrounding area are obtained; The distance and the numerical difference between any two of the maximum value points are analyzed to obtain the numerical characteristics of the projected data points in the projection plane.

[0011] In some embodiments of the present invention, analyzing the positional relationship between the superpixel clusters in the cluster to obtain the distribution factor of the cluster includes: Obtaining the centroid of each superpixel cluster in the cluster; In the clustering, obtaining the minimum distance between the centroid of each superpixel cluster and the centroids of all other superpixel clusters, recorded as the third distance of the superpixel cluster; In the clustering, analyzing the information entropy of the position distribution between each of the superpixel clusters and all other superpixel clusters; A distribution factor of the cluster is obtained according to the third distance and the information entropy.

[0012] In some embodiments of the present invention, in the clustering, analyzing the information entropy of the position distribution between each superpixel cluster and all other superpixel clusters includes: Extending in a horizontal direction through the centroid of the superpixel cluster to obtain an extension line of the superpixel cluster; In the clustering, the centroids of the superpixel cluster and each other superpixel cluster are connected to obtain the length of the centroid connection line, which is recorded as the fourth distance, and the angle of the clockwise angle between the centroid connection line and the extension line is obtained, which is recorded as the first angle; The product of the fourth distance between the superpixel cluster and each other superpixel cluster and the first angle is calculated to obtain the information entropy of the superpixel cluster.

[0013] In some embodiments of the present invention, obtaining the distribution factor of the cluster according to the third distance and the information entropy includes: The discrete degrees between the third distances corresponding to all the superpixel clusters in the cluster are analyzed, and the discrete degrees between the information entropies corresponding to all the superpixel clusters in the cluster are analyzed to obtain the distribution factor of the cluster.

[0014] In some embodiments of the present invention, analyzing the volume difference and volume distribution of the superpixel clusters in the cluster to obtain the noise impact factor of the cluster includes: Obtaining the superpixel cluster with the largest volume in the cluster, recorded as the first cluster; Calculating the average volume ratio of all remaining superpixel clusters in the cluster to the first cluster to obtain the volume difference of the superpixel clusters in the cluster; Obtaining two superpixel clusters that are closest to each superpixel cluster in the cluster and the superpixel clusters that are closest to the superpixel cluster in the cluster, and recording them as close clusters; Connect the centroid of the superpixel cluster to the centroids of two corresponding close clusters, and record the angle between the two centroids as the second angle; Analyzing the difference between the second angle of each superpixel cluster in the cluster and the second angles of other superpixel clusters, and combining the volume of the superpixel clusters to obtain the volume distribution of the superpixel clusters in the cluster; The noise impact factor of the clustering is obtained by combining the volume difference and the volume distribution.

[0015] In some embodiments of the present invention, according to the noise impact degree, the spatial point cloud data is subjected to noise reduction processing, including: The spatial point cloud data is subjected to noise reduction processing based on a voxel grid filtering algorithm, and according to the degree of influence of the noise corresponding to each projection plane, the voxel size in the voxel grid filtering algorithm is adjusted to perform noise reduction processing on the spatial point cloud data corresponding to each projection plane.

[0016] Compared with the prior art, the method for constructing a three-dimensional model of building structure design based on spatial point cloud data provided by the present invention has the following beneficial effects: The present invention collects spatial point cloud data, first divides the spatial point cloud data into several projection planes according to the construction building, then divides the superpixel clusters according to the distribution characteristics and numerical characteristics of the projection data points in the projection plane, and obtains the clustering of superpixel clusters by analyzing the similarities between different superpixel clusters, realizes the functional block segmentation of the projection plane, facilitates the subsequent accurate analysis of noise impact, and then performs accurate and effective noise reduction processing; then, by analyzing the positional relationship between superpixel clusters in the cluster, and analyzing the volume difference and volume distribution of superpixel clusters in the cluster, and then combining the volume of the cluster, the noise impact degree of the projection plane is obtained; finally, according to the noise impact degree, the spatial point cloud data is subjected to noise reduction processing, which provides good data support for the subsequent construction of the three-dimensional model of the construction building. The method of the present invention improves the efficiency and accuracy of the noise reduction processing by performing different degrees of noise reduction processing on the spatial point cloud data of different projection planes and different clusters of the construction building, thereby helping to build a reliable three-dimensional model of building structure design. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. 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 creative work.

[0018] Figure 1A schematic diagram of the basic process of a method for constructing a three-dimensional model of a building structure design based on spatial point cloud data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the method for constructing a three-dimensional model of architectural structure design based on spatial point cloud data proposed by the present invention, its specific implementation method, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art of the present invention. Terms such as "comprises", "comprising" or any other variant thereof are intended to cover non-exclusive inclusions, so that a circuit structure, 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 articles or devices. In the absence of further restrictions, there are elements defined by the statement "comprising one...", and it is not excluded that there are other identical elements in the articles or devices including the elements. Relational terms such as the terms "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.

[0021] It should be noted that in order to ensure that the calculation results are meaningful, when performing fractional operations in the embodiments of the present invention, when the denominator is 0, it is necessary to add a parameter adjustment factor greater than 0 to the denominator to prevent the denominator from being 0. The value of the parameter adjustment factor is set by the implementer according to actual conditions, and this application does not impose any special restrictions.

[0022] The following is a detailed description of a method for constructing a three-dimensional model of a building structure design based on spatial point cloud data provided by the present invention in conjunction with the accompanying drawings.

[0023] See also Figure 1 , which shows the basic process of a method for constructing a three-dimensional model of architectural structure design based on spatial point cloud data provided by an embodiment of the present invention.

[0024] like Figure 1 As shown, a method for constructing a three-dimensional model of a building structure design based on spatial point cloud data provided by an embodiment of the present invention specifically includes: S100: Collect spatial point cloud data of the building.

[0025] In order to construct a three-dimensional model of the construction building, it is first necessary to collect the spatial point cloud data of the construction building. Due to the large volume of the construction building, it is difficult to collect spatial point cloud data with a handheld laser scanner, so the spatial point cloud data is collected by mounting a laser scanner on a drone. Specifically, first select a laser scanner with a suitable scanning range and accuracy according to the volume of the construction building, and select a waterproof and dustproof laser scanner to adapt to the harsh construction environment. Calibrate the laser scanner before use to ensure accurate measurement; then install the laser scanner on the drone, and set the flight trajectory of the drone to continuously circle the construction building; the laser scanner emits a laser beam, and the laser beam is reflected back to the scanner after hitting the target surface. The equipment calculates the distance from the target surface point to the laser scanner based on the return time of the laser beam, and records the three-dimensional coordinates of the target surface point; during the scanning process of the laser scanner, the laser beam is continuously emitted to generate a large amount of point data, each point contains spatial coordinate information, and these points converge into a spatial point cloud data set for the construction building.

[0026] At this point, the spatial point cloud data of the construction building has been obtained.

[0027] S200: Divide the spatial point cloud data into data points of different grid surfaces, project the data points in each grid surface, and obtain a projection plane corresponding to each grid surface.

[0028] Since the spatial point cloud data reflects the three-dimensional structure of the construction building, and the distribution of data points on each surface of the construction building is not the same, it is necessary to divide the spatial point cloud data into data points of different grid surfaces, that is, to divide the spatial point cloud data into different surfaces of the construction building. Specifically, firstly, the surface of the spatial point cloud data is reconstructed using the triangular mesh model, and the mesh in the triangular mesh algorithm is divided into multiple mesh surfaces, and then the regional growing algorithm is used to segment the mesh surface, and the spatial point cloud data is divided into data points of different mesh surfaces.

[0029] The data points of different grid surfaces obtained represent the data points of the construction building at different azimuth viewing angles. Therefore, further, the data points in each grid surface are projected. Specifically, all data points in each grid surface are projected to a plane perpendicular to its viewing angle to obtain a projection plane corresponding to each grid surface, wherein the data value of the projected data point in the projection plane is the height value between the data point corresponding to the projection data point and the projection plane, and each data point in the projection plane is discrete.

[0030] At this point, all spatial point cloud data are projected onto the corresponding projection plane, and the projection data points in the projection plane are obtained.

[0031] During the acquisition process of spatial point cloud data, due to the accuracy of the laser scanner, and the abnormal points or noise points generated during the transmission of spatial point cloud data, and due to the irregular distribution of the building structure, different parts are affected by noise differently. Therefore, it is necessary to divide the projection plane into different parts and analyze the noise conditions of different parts, which specifically includes steps S300 to S700.

[0032] S300: Analyze the distribution characteristics and numerical characteristics of the projection data points in the projection plane, perform superpixel segmentation on the projection data points in the projection plane, and obtain multiple superpixel clusters.

[0033] There are various functional blocks in different areas of the construction building. The characteristics of the spatial point cloud data of different functional blocks are different. Therefore, before performing noise analysis, all projection planes need to be segmented into functional blocks.

[0034] Therefore, in an embodiment of the present invention, by analyzing the distribution characteristics and numerical characteristics of the projection data points in the projection plane, the projection data points in the projection plane are segmented into superpixels to obtain multiple superpixel clusters. The distribution characteristics of the projected data points in the projection plane are analyzed. The specific implementation method is as follows: first, in the projection plane, the first minimum distance between all the projected data points and other projected data points is obtained, and the maximum value of the first minimum distances corresponding to all the projected data points is recorded as the first distance; then, for each projected data point in the projection plane, a circle is drawn in the projection plane with the projection data point as the center and the first distance as the radius to obtain the surrounding area of ​​each projected data point; finally, the density of the projected data points in the surrounding area corresponding to each projected data point is analyzed, that is, the ratio of the number of projected data points in the surrounding area to the area of ​​the surrounding area (the area of ​​the circle with the radius of the first distance) is calculated to obtain the distribution characteristics of the projected data points in the projection plane.

[0035] The numerical characteristics of the projected data points in the projection plane are analyzed. The specific implementation method is as follows: first, in the surrounding area of ​​the projection data point, the second minimum distance between all the projection data points and other projection data points is obtained, and the maximum value of the second minimum distances corresponding to all the projection data points is recorded as the second distance; then, for any projection data point in the surrounding area, if its data value is the largest within the second distance range (with the projection data point as the center and the second distance range as the radius), the projection data point is recorded as a maximum point, and all the maximum points in the surrounding area corresponding to each projection data point are obtained; finally, the distance and numerical difference between any two maximum points in each surrounding area are analyzed to obtain the numerical characteristics of the projection data points in the projection plane.

[0036] The projected data points on the projection plane are segmented into superpixels to obtain a plurality of superpixel clusters. The specific implementation method is as follows: according to the distribution characteristics of the projected data points in the projection plane (the density of the projected data points in the surrounding area) and the numerical characteristics of the projected data points in the projection plane, the projected data points on the projection plane are segmented into superpixels to obtain a plurality of superpixel clusters, and the distribution characteristics and numerical characteristics of the projected data points within each superpixel cluster are similar.

[0037] S400: Analyze the similarities between superpixel clusters based on the distribution characteristics and numerical characteristics of the projected data points inside the superpixel clusters to obtain clustering of the superpixel clusters.

[0038] Different functional areas on the surface of the construction building will have different densities and fluctuations in the spatial point clouds collected from different parts due to different structures and materials. Some smaller superpixel clusters may be caused by noise and some special structures on the surface of the construction building. Therefore, it is necessary to analyze the distribution of superpixel clusters in each projection plane of the spatial point cloud data of the construction building.

[0039] In an embodiment of the present invention, based on the distribution characteristics and numerical characteristics of the projected data points inside the superpixel clusters, the similarities between the superpixel clusters are analyzed to obtain the clustering of the superpixel clusters. The specific implementation method is: for the superpixel clusters in the acquired projection plane, the numerical characteristics, distribution characteristics (density) and the two superpixel clusters with the closest data values ​​are obtained and recorded as similar clusters. For the similar clusters of superpixel clusters, if some superpixel clusters belong to similar clusters of several superpixel clusters, the superpixel cluster and the superpixel cluster that is most similar to it are recorded as a pair of similar clusters. In order to ensure that the superpixel clusters in the cluster are relatively similar, the similar clusters obtained by the above operation, if the remaining superpixel clusters and any superpixel cluster in the similar clusters belong to similar clusters, they are divided into the similar clusters, thereby obtaining the clustering of several similar clusters, that is, obtaining the clustering of pairs of superpixel clusters, and the internal projection data point distribution of the superpixel clusters in the same cluster is relatively similar.

[0040] S500: Analyze the positional relationship between superpixel clusters in the cluster to obtain the distribution factor of the cluster.

[0041] The clustering of superpixel clusters obtained in the above steps, the numerical characteristics, distribution characteristics (density) and data values ​​of the projected data points between superpixel clusters in the same cluster are relatively similar. However, the position distribution between different superpixel clusters in the same cluster will affect the distribution regularity of superpixel clusters within the entire cluster, and the distribution regularity of superpixel clusters is an important manifestation of the influence of noise.

[0042] Therefore, by analyzing the positional relationship between superpixel clusters in the cluster, the distribution factor of the cluster is obtained. The specific implementation method is: first, obtain the centroid of each superpixel cluster in the cluster; then, in the cluster, obtain the minimum distance between the centroid of each superpixel cluster and the centroid of all other superpixel clusters, recorded as the third distance of the superpixel cluster; in the cluster, analyze the information entropy of the position distribution between each superpixel cluster and all other superpixel clusters. A more specific implementation method is to extend the centroid of the superpixel cluster in the horizontal direction to obtain the extension line of the superpixel cluster; in the cluster, connect the centroids between the superpixel cluster and each other superpixel cluster to obtain the length of the centroid connection line, recorded as the fourth distance (each superpixel cluster corresponds to The fourth distance, represents the number of superpixel clusters in the cluster), and obtains the clockwise angle between the centroid line and the extension line, which is recorded as the first angle (each superpixel cluster corresponds to first angle); calculate the product of the fourth distance and the first angle between the superpixel cluster and each other superpixel cluster (each superpixel cluster corresponds to The information entropy of the superpixel clusters is obtained according to the product; finally, the distribution factor of the cluster is obtained according to the third distance and the information entropy. A more specific implementation method is to analyze the discrete degree between the third distances corresponding to all superpixel clusters in the cluster and the discrete degree between the information entropies corresponding to all superpixel clusters in the cluster to obtain the distribution factor of the cluster. Construct the first The distribution factor calculation formula for a cluster is: In the formula, Represents the projection plane The distribution factor of the clusters; Indicates In the cluster The information entropy of superpixel clusters; Indicates The variance of the information entropy corresponding to all superpixel clusters in a cluster; Indicates In the cluster The third distance of superpixel clusters; Indicates The variance of the third distances corresponding to all superpixel clusters in a cluster; Indicated by natural constant An exponential function with base .

[0043] The variance of the information entropy corresponding to all superpixel clusters in the cluster The smaller the value, the smaller the difference in information entropy between superpixel clusters in the cluster, indicating that the fourth distance and the first angle of the superpixel clusters in the cluster are not much different, indicating that the distribution of superpixel clusters in the cluster is similar, so the distribution factor of the cluster is larger; for all superpixel clusters in the cluster, the variance of the third distance corresponding to The smaller the value, the more evenly the distance distribution between superpixel clusters in the cluster is distributed, so the distribution factor of the cluster is larger; the larger the distribution factor, the more regular the distribution of superpixel clusters in the cluster.

[0044] S600: Analyze the volume difference and volume distribution of superpixel clusters in the cluster to obtain the noise impact factor of the cluster.

[0045] Since the distribution of noise in the projection plane of the construction building is disordered, it may be affected by some noise in the cluster, resulting in a lower distribution factor of the cluster. The noise will be reflected in the cluster as a superpixel cluster with a smaller volume, where the volume is the number of projection data points, and there is a large difference between it and other superpixel clusters in the cluster. Therefore, in an embodiment of the present invention, the noise influencing factor of the cluster is obtained by analyzing the volume difference and volume distribution of the superpixel clusters in the cluster. The specific implementation method is as follows: first, obtain the superpixel cluster with the largest volume in the cluster, recorded as the first cluster; and calculate the average volume ratio of all remaining superpixel clusters in the cluster to the first cluster, and obtain the volume difference corresponding to all superpixel clusters in the cluster; then, obtain the two superpixel clusters closest to the volume of each superpixel cluster in the cluster, recorded as close clusters; and connect the centroid of the superpixel cluster with the centroids of its two corresponding close clusters respectively, and record the angle between the two centroids as the second angle; then, analyze the difference between the second angle of each superpixel cluster in the cluster and the second angles of other superpixel clusters, that is, calculate the difference between the second angle of each superpixel cluster and the mean second angle of all superpixel clusters in the cluster, and combine the volume of superpixel clusters to obtain the volume distribution of superpixel clusters in the cluster; finally, combine the volume difference and volume distribution to obtain the noise influencing factor of the cluster. Construct the first The calculation formula of the noise impact factor of a cluster is: In the formula, Represents the projection plane The noise impact factor of each cluster; Indicates The average volume ratio of all remaining superpixel clusters in the cluster to the first cluster; Indicates In the cluster The second angle corresponding to the superpixel clusters; Indicates The mean of the second angles corresponding to all superpixel clusters in the cluster; Indicates In the cluster The volume size of the superpixel cluster; Indicates The number of superpixel clusters in a cluster; Represents a hyperbolic function.

[0046] The average volume ratio for clusters , represents the volume difference of all superpixel clusters in the cluster. The larger its value, the closer the volume of superpixel clusters in the cluster. Therefore, when a superpixel cluster with a smaller volume appears, it may be affected by noise. For the superpixel cluster in the cluster, the ratio of the difference between the second angle and the mean of the second angle to the volume of the superpixel cluster is The larger the value, the larger the angle between superpixel clusters with similar volumes and the smaller the volume of the superpixel cluster, which means that the cluster is more affected by noise.

[0047] S700: Obtain the noise impact degree of the projection plane according to the distribution factor and the noise impact factor of the cluster and the volume of the cluster.

[0048] Through the above steps, the distribution factor and noise impact factor of each cluster in the projection plane are obtained, and the noise impact on the overall projection plane needs to be obtained. Therefore, in an embodiment of the present invention, the noise impact degree of the projection plane is obtained according to the distribution factor and noise impact factor of the cluster, combined with the volume of the cluster. The larger the volume of the cluster, the lower the distribution factor and the larger the noise impact factor, the higher the overall noise impact on the projection plane. Therefore, the calculation formula for the noise impact degree of the projection plane is constructed as follows: In the formula, Indicates the degree of noise influence on the projection plane; represents the number of clusters in the projection plane; Represents the projection plane The distribution factor of the clusters; Represents the projection plane The noise impact factor of each cluster; Represents the projection plane The volume of a cluster.

[0049] S800: Perform noise reduction processing on spatial point cloud data according to the degree of noise impact.

[0050] According to the degree of noise influence, the spatial point cloud data is subjected to noise reduction processing. The specific implementation method is as follows: Voxel grid filtering is a common method for denoising spatial point cloud data. The adjustment of voxel size in the voxel grid filtering algorithm is a key step in voxel grid filtering denoising, because the size of voxels directly affects the denoising effect and detail retention of spatial point cloud data. Therefore, the spatial point cloud data is denoised based on the voxel grid filtering algorithm, and the voxel size in the voxel grid filtering algorithm is adjusted according to the degree of noise influence corresponding to each projection plane. The specific method of voxel adjustment is as follows: first, the initial voxel size is selected according to the volume and density of the spatial point cloud data. In the initial voxel case, the spatial point cloud data will be relatively rough and need to be filtered and smoothed. Due to the complexity of the construction environment, the spatial point cloud data in different parts have different degrees of noise influence and thus have different degrees of variation. It is necessary to adjust the voxel size according to the noise influence factor of different projection planes. Therefore, the voxel size is adjusted to: In the formula, represents the adjusted voxel size; represents the initial voxel size; Indicates The degree of noise influence on the projection plane.

[0051] According to the adjusted voxel size, the spatial point cloud data corresponding to each projection plane is denoised.

[0052] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0053] Various embodiments in this specification are described in a progressive manner, and the same or similar parts between various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for constructing a three-dimensional model of building structure design based on spatial point cloud data, characterized in that: The method comprises: Collect spatial point cloud data of buildings; Dividing the spatial point cloud data into data points of different grid surfaces, projecting the data points in each grid surface, and obtaining a projection plane corresponding to each grid surface; Analyzing the distribution characteristics and numerical characteristics of the projection data points in the projection plane, performing superpixel segmentation on the projection data points in the projection plane, and obtaining a plurality of superpixel clusters; Analyzing the similarities between the superpixel clusters based on the distribution characteristics and numerical characteristics of the projection data points inside the superpixel clusters to obtain clustering of the superpixel clusters; Analyzing the positional relationship between the superpixel clusters in the cluster to obtain a distribution factor of the cluster; Analyzing the volume difference and volume distribution of the superpixel clusters in the cluster to obtain the noise impact factor of the cluster; According to the distribution factor and the noise impact factor of the cluster, combined with the volume of the cluster, the noise impact degree of the projection plane is obtained; According to the noise impact degree, noise reduction processing is performed on the spatial point cloud data.

2. The method for constructing a three-dimensional model of a building structure design based on spatial point cloud data according to claim 1, characterized in that: Dividing the spatial point cloud data into data points of different grid surfaces includes: The spatial point cloud data is surface reconstructed using a triangular mesh model, and the mesh in the triangular mesh algorithm is divided into a plurality of mesh surfaces, and then the mesh surfaces are segmented using a region growing algorithm to divide the spatial point cloud data into data points on different mesh surfaces.

3. The method for constructing a three-dimensional model of a building structure design based on spatial point cloud data according to claim 2, characterized in that: Project the data points in each mesh surface to obtain the projection plane corresponding to each mesh surface, including: Project all data points in each grid surface onto a plane perpendicular to its viewing angle to obtain a projection plane corresponding to each grid surface, wherein the data value of the projected data point in the projection plane is the height value between the data point corresponding to the projection data point and the projection plane.

4. The method for constructing a three-dimensional model of a building structure design based on spatial point cloud data according to claim 3, characterized in that: Analyzing the distribution characteristics of the projection data points in the projection plane includes: In the projection plane, obtaining the first minimum distance between all projection data points and other projection data points, and recording the maximum value of all first minimum distances as the first distance; For each projection data point in the projection plane, a circle is drawn in the projection plane with the projection data point as the center and the first distance as the radius to obtain the surrounding area of ​​the projection data point; The density of the projected data points in the surrounding area is analyzed to obtain distribution characteristics of the projected data points in the projection plane.

5. The method for constructing a three-dimensional model of building structure design based on spatial point cloud data according to claim 4, characterized in that: Analyzing the numerical characteristics of the projected data points in the projection plane includes: In the surrounding area, obtaining the second minimum distances between all projection data points and other projection data points, and recording the maximum value of all the second minimum distances as the second distance; For any projected data point in the surrounding area, if its data value is the largest within the second distance range, the projected data point is recorded as a maximum value point, and all maximum value points in the surrounding area are obtained; The distance and the numerical difference between any two of the maximum value points are analyzed to obtain the numerical characteristics of the projected data points in the projection plane.

6. The method for constructing a three-dimensional model of building structure design based on spatial point cloud data according to claim 1, characterized in that: Analyzing the positional relationship between the superpixel clusters in the cluster to obtain the distribution factor of the cluster includes: Obtaining the centroid of each superpixel cluster in the cluster; In the clustering, obtaining the minimum distance between the centroid of each superpixel cluster and the centroids of all other superpixel clusters, recorded as the third distance of the superpixel cluster; In the clustering, analyzing the information entropy of the position distribution between each of the superpixel clusters and all other superpixel clusters; A distribution factor of the cluster is obtained according to the third distance and the information entropy.

7. The method for constructing a three-dimensional model of a building structure design based on spatial point cloud data according to claim 6, characterized in that: In the clustering, analyzing the information entropy of the position distribution between each superpixel cluster and all other superpixel clusters, including: Extending in a horizontal direction through the centroid of the superpixel cluster to obtain an extension line of the superpixel cluster; In the clustering, the centroids of the superpixel cluster and each other superpixel cluster are connected to obtain the length of the centroid connection line, which is recorded as the fourth distance, and the angle of the clockwise angle between the centroid connection line and the extension line is obtained, which is recorded as the first angle; The product of the fourth distance between the superpixel cluster and each other superpixel cluster and the first angle is calculated to obtain the information entropy of the superpixel cluster.

8. The method for constructing a three-dimensional model of a building structure design based on spatial point cloud data according to claim 6, characterized in that: Obtaining a distribution factor of the cluster according to the third distance and the information entropy includes: The discrete degrees between the third distances corresponding to all the superpixel clusters in the cluster are analyzed, and the discrete degrees between the information entropies corresponding to all the superpixel clusters in the cluster are analyzed to obtain the distribution factor of the cluster.

9. The method for constructing a three-dimensional model of building structure design based on spatial point cloud data according to claim 1, characterized in that: Analyze the volume difference and volume distribution of the superpixel clusters in the cluster to obtain the noise impact factor of the cluster, including: Obtaining the superpixel cluster with the largest volume in the cluster, recorded as the first cluster; Calculating the average volume ratio of all remaining superpixel clusters in the cluster to the first cluster to obtain the volume difference of the superpixel clusters in the cluster; Obtaining two superpixel clusters that are closest to each superpixel cluster in the cluster and the superpixel clusters that are closest to the superpixel cluster in the cluster, and recording them as close clusters; Connect the centroid of the superpixel cluster to the centroids of two corresponding close clusters, and record the angle between the two centroids as the second angle; Analyzing the difference between the second angle of each superpixel cluster in the cluster and the second angles of other superpixel clusters, and combining the volume of the superpixel clusters to obtain the volume distribution of the superpixel clusters in the cluster; The noise impact factor of the clustering is obtained by combining the volume difference and the volume distribution.

10. The method for constructing a three-dimensional model of building structure design based on spatial point cloud data according to claim 1, characterized in that: According to the noise impact degree, the spatial point cloud data is subjected to noise reduction processing, including: The spatial point cloud data is subjected to noise reduction processing based on a voxel grid filtering algorithm, and according to the degree of influence of the noise corresponding to each projection plane, the voxel size in the voxel grid filtering algorithm is adjusted to perform noise reduction processing on the spatial point cloud data corresponding to each projection plane.

Images