Construction quality acceptance method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of building quality monitoring, and particularly provides a construction quality acceptance method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the point cloud data of a construction site, obtaining a building information model corresponding to the construction site, carrying out the noise reduction of the point cloud data, obtaining the noise reduction point cloud data, and storing the noise reduction point cloud data in a database; and performing down-sampling processing on the noise reduction point cloud data and the building information model to obtain down-sampled noise reduction point cloud data and down-sampled model point cloud data, and performing point cloud registration on the down-sampled noise reduction point cloud data and the down-sampled model point cloud data through an FREAK algorithm and an ICP algorithm to obtain the noise reduction point cloud data and the model point cloud data. And obtaining construction quality acceptance result information of the construction site. And the construction site is checked by using the building information model, the FREAK algorithm and the ICP algorithm, so that the project acceptance efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of building quality monitoring, and in particular to a construction quality acceptance method, device, electronic equipment and storage medium. Background Art

[0002] In the field of construction project quality management, construction inspection and acceptance are key to ensuring project quality. With the advancement of digitalization in the construction industry, an increasing number of digital products and technologies are being applied to this field, such as mobile supervision apps, cloud platforms and big data, and intelligent monitoring and early warning systems. Despite the application of numerous digital products in this field, manual inspections have long been inefficient due to the complex construction environment and content. The lack of intelligent auxiliary acceptance and inspection methods has become a pain point in the industry. Traditional construction progress monitoring relies primarily on manual on-site inspection and record-keeping. However, project construction sites are rich in data and contain a large amount of information, making manual collection time-consuming and labor-intensive.

[0003] With the rapid development of Building Information Modeling (BIM) and 3D scanning technologies, intelligent inspection systems based on point cloud data have revolutionized construction acceptance. While existing methods exist for comparing BIM models with on-site scan data, practical applications still face the following technical bottlenecks: 1) Complex construction environments, with issues such as personnel movement and equipment obstruction, can lead to missing point cloud data; 2) Traditional ICP registration algorithms are computationally inefficient and time-consuming; and 3) a lack of registration optimization methods tailored to the specific construction scenarios results in insufficient comparison accuracy.

[0004] Therefore, it is of great significance to study a construction quality acceptance method based on building information model, FREAK algorithm and ICP algorithm to quickly and accurately inspect the construction site using building information model, FREAK algorithm and ICP algorithm, which is of great significance for project acceptance, ensuring project acceptance and accuracy, and improving the efficiency of project acceptance. Summary of the Invention

[0005] The present invention provides a construction quality acceptance method, device, electronic device and storage medium, which are used to solve the problems of missing point cloud data, low calculation efficiency and insufficient registration accuracy in the existing acceptance method of comparing BIM models with on-site scan data, laying an important foundation for the accurate acceptance of construction quality.

[0006] In a first aspect, the present invention provides a construction quality acceptance method, comprising: Acquiring on-site point cloud data of a construction site and acquiring a building information model corresponding to the construction site; Performing noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data; Downsampling the noise-reduced point cloud data and the building information model to obtain downsampled noise-reduced point cloud data and downsampled model point cloud data; The downsampled denoised point cloud data and the downsampled model point cloud data are aligned using the FREAK algorithm and the ICP algorithm to obtain construction quality acceptance result information of the construction site.

[0007] Optionally, performing noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data includes: The average distance between each point cloud and the neighboring points in the on-site point cloud data is calculated by the K-nearest neighbor algorithm; Calculating a noise reduction threshold based on the average distance using a Gaussian distribution and a preset noise reduction threshold calculation formula; Point clouds with an average distance greater than the noise reduction threshold are removed from the on-site point cloud data to obtain noise-reduced point cloud data.

[0008] Optionally, the noise reduction threshold is calculated based on the average distance using a Gaussian distribution and a preset noise reduction threshold calculation formula, including: Using Gaussian distribution, calculate the mean and standard deviation of the average distance; The mean and the standard deviation are input into a preset noise reduction threshold calculation formula to calculate the noise reduction threshold.

[0009] Optionally, downsampling the de-noised point cloud data and the building information model to obtain downsampled de-noised point cloud data and downsampled model point cloud data includes: Discretizing the building information model to obtain discrete point cloud data corresponding to the building information model; resampling the discrete point cloud data according to the noise reduction point cloud data to obtain resampled discrete point cloud data; A density-adaptive voxel grid downsampling method is used to downsample the resampled discrete point cloud data and the denoised point cloud data to obtain downsampled denoised point cloud data and downsampled model point cloud data.

[0010] Optionally, resampling the discrete point cloud data according to the noise reduction point cloud data to obtain resampled discrete point cloud data includes: Calculating the average spacing of each point cloud in the noise-reduced point cloud data; The discrete point cloud data is resampled using the average spacing to obtain resampled discrete point cloud data.

[0011] Optionally, a density-adaptive voxel grid downsampling method is used to downsample the resampled discrete point cloud data and the de-noised point cloud data to obtain downsampled de-noised point cloud data and downsampled model point cloud data, including: Based on a preset voxel side length, dividing the resampled discrete point cloud data and the de-noised point cloud data into a plurality of three-dimensional voxel grids, respectively, to obtain a plurality of first three-dimensional voxel grid data corresponding to the resampled discrete point cloud data and a plurality of second three-dimensional voxel grid data corresponding to the de-noised point cloud data; respectively determining whether the number of point clouds in each of the first three-dimensional voxel grid data and each of the second three-dimensional voxel grid data is greater than or equal to a preset minimum number of voxel points; if so, determining the point cloud closest to the center point in the first three-dimensional voxel grid data or the second three-dimensional voxel grid data whose number of point clouds is greater than or equal to the preset minimum number of voxel points as a representative point cloud; if not, adjusting the voxel side length until the number of point clouds in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is greater than or equal to the preset minimum number of voxel points, and determining the point cloud closest to the center point in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data as the representative point cloud; Point clouds other than the representative point cloud are eliminated from the resampled discrete point cloud data and the denoised point cloud data to obtain downsampled denoised point cloud data and downsampled model point cloud data.

[0012] Optionally, point cloud registration is performed on the downsampled denoised point cloud data and the downsampled model point cloud data using a FREAK algorithm and an ICP algorithm to obtain construction quality acceptance result information of the construction site, including: Comparing the downsampled denoised point cloud data with actual site information of the construction site, and repairing missing data in the downsampled denoised point cloud data using a depth completion algorithm to obtain repaired denoised point cloud data; performing coarse registration on the repaired denoised point cloud data and the downsampled model point cloud data by a feature point matching method, so as to align point clouds at corresponding positions in the repaired denoised point cloud data and the downsampled model point cloud data; The FREAK algorithm and the ICP algorithm are used to perform precise registration on the aligned repaired denoised point cloud data and the downsampled model point cloud data to obtain construction quality acceptance result information of the construction site.

[0013] In a second aspect, the present invention provides a construction quality acceptance device, comprising: An acquisition module, configured to acquire on-site point cloud data of a construction site and a building information model corresponding to the construction site; A noise reduction module, configured to perform noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data; a downsampling module, configured to perform downsampling processing on the de-noised point cloud data and the building information model to obtain downsampled de-noised point cloud data and downsampled model point cloud data; The registration module is used to perform point cloud registration on the downsampled denoised point cloud data and the downsampled model point cloud data through the FREAK algorithm and the ICP algorithm to obtain the construction quality acceptance result information of the construction site.

[0014] In a third aspect, the present invention provides an electronic device comprising a processor and a memory, wherein the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are executed.

[0015] In a fourth aspect, the present invention provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, runs the steps of the method provided in the first aspect above.

[0016] It can be seen from the above technical solutions that the present invention has the following advantages: The present invention provides a construction quality acceptance method, device, electronic device, and storage medium. The method comprises: obtaining on-site point cloud data of a construction site, obtaining a building information model corresponding to the construction site, performing noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data, downsampling the noise-reduced point cloud data and the building information model to obtain noise-reduced point cloud data and model point cloud data after downsampling, and performing point cloud registration on the noise-reduced point cloud data and the model point cloud data after downsampling using a FREAK algorithm and an ICP algorithm to obtain construction quality acceptance result information for the construction site. This method forms a set of methods for inspecting construction sites using a building information model, a FREAK algorithm, and an ICP algorithm, addressing the problems of missing point cloud data, low computational efficiency, and insufficient registration accuracy in existing acceptance methods that compare BIM models with on-site scan data, thereby laying an important foundation for accurate construction quality acceptance. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 This is a flowchart of a construction quality acceptance method according to a first embodiment of the present invention; Figure 2 This is a flow chart of a second embodiment of a construction quality acceptance method of the present invention; Figure 3 This is a structural block diagram of an embodiment of a construction quality acceptance device of the present invention. DETAILED DESCRIPTION

[0019] Embodiments of the present invention provide a construction quality acceptance method, device, electronic device and storage medium, which are used to solve the problems of missing point cloud data, low calculation efficiency and insufficient registration accuracy in the existing acceptance method of comparing BIM models with on-site scan data, laying an important foundation for the accurate acceptance of construction quality.

[0020] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Example 1

[0021] See also Figure 1 , Figure 1 This is a flow chart of a construction quality acceptance method according to a first embodiment of the present invention, the method comprising: Step S101, obtaining on-site point cloud data of a construction site, and obtaining a building information model corresponding to the construction site; The embodiment of the present invention obtains on-site point cloud data of the construction site through a camera (or equipment such as a drone or robot that can be equipped with a camera), and obtains the building information model corresponding to the construction site from a database (such as a BIM collaborative management platform).

[0022] Step S102, performing noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data; In an optional embodiment, performing noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data includes: The average distance between each point cloud and the neighboring points in the on-site point cloud data is calculated by the K-nearest neighbor algorithm; Calculating a noise reduction threshold based on the average distance using a Gaussian distribution and a preset noise reduction threshold calculation formula; Point clouds with an average distance greater than the noise reduction threshold are removed from the on-site point cloud data to obtain noise-reduced point cloud data.

[0023] In the embodiment of the present invention, the K-nearest neighbor algorithm is used to calculate the average distance between the corresponding neighborhood points of each point cloud in the point cloud data. The mean and standard deviation of the average distance are calculated using the Gaussian distribution. The mean and standard deviation are input into a preset noise reduction threshold calculation formula to calculate the noise reduction threshold. Point clouds with an average distance greater than the noise reduction threshold are removed from the point cloud data to obtain noise-reduced point cloud data.

[0024] Step S103, downsampling the noise-reduced point cloud data and the building information model to obtain downsampled noise-reduced point cloud data and downsampled model point cloud data; In an optional embodiment, downsampling the de-noised point cloud data and the building information model to obtain downsampled de-noised point cloud data and downsampled model point cloud data includes: Discretizing the building information model to obtain discrete point cloud data corresponding to the building information model; resampling the discrete point cloud data according to the noise reduction point cloud data to obtain resampled discrete point cloud data; A density-adaptive voxel grid downsampling method is used to downsample the resampled discrete point cloud data and the denoised point cloud data to obtain downsampled denoised point cloud data and downsampled model point cloud data.

[0025] The embodiment of the present invention discretizes a building information model to obtain discrete point cloud data corresponding to the building information model, performs point cloud registration on the downsampled denoised point cloud data and the downsampled model point cloud data through the FREAK algorithm and the ICP algorithm, obtains construction quality acceptance result information of the construction site, calculates the average spacing of each point cloud in the denoised point cloud data, uses the average spacing to resample the discrete point cloud data to obtain resampled discrete point cloud data, divides the resampled discrete point cloud data and the denoised point cloud data into a plurality of three-dimensional voxel grids based on a preset voxel side length, obtains a plurality of first three-dimensional voxel grid data corresponding to the resampled discrete point cloud data and a plurality of second three-dimensional voxel grid data corresponding to the denoised point cloud data, and judges the size of each first three-dimensional voxel grid data and Whether the number of point clouds in each second three-dimensional voxel grid data is greater than or equal to the preset minimum number of voxel points; if so, the point cloud closest to the center point in the first three-dimensional voxel grid data or the second three-dimensional voxel grid data whose point cloud number is greater than or equal to the preset minimum number of voxel points is determined as the representative point cloud; if not, the voxel side length is adjusted until the number of point clouds in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is greater than or equal to the preset minimum number of voxel points, and the point cloud closest to the center point in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is determined as the representative point cloud, and the point clouds other than the representative point cloud in the resampled discrete point cloud data and the denoised point cloud data are eliminated to obtain the downsampled denoised point cloud data and the downsampled model point cloud data.

[0026] Step S104: performing point cloud registration on the downsampled noise reduction point cloud data and the downsampled model point cloud data using the FREAK algorithm and the ICP algorithm to obtain a construction quality acceptance result of the construction site; An embodiment of the present invention compares the denoised point cloud data after downsampling and the model point cloud data after downsampling, and uses a depth completion algorithm to repair missing data in the denoised point cloud data after downsampling to obtain repaired denoised point cloud data. The repaired denoised point cloud data and the downsampling model point cloud data are roughly aligned through a feature point matching method to align point clouds at corresponding positions in the repaired denoised point cloud data and the downsampling model point cloud data. The FREAK algorithm and the ICP algorithm are used to finely align the aligned repaired denoised point cloud data and the downsampling model point cloud data to obtain construction quality acceptance result information of the construction site.

[0027] The embodiment of the present invention provides a construction quality acceptance method that uses a building information model, a FREAK algorithm, and an ICP algorithm to inspect a construction site, thereby improving the efficiency of project acceptance and laying an important foundation for accurate acceptance of construction quality. Example 2

[0028] See also Figure 2 , Figure 2 This is a flow chart of a second embodiment of a construction quality acceptance method of the present invention, the steps comprising: S201, obtaining on-site point cloud data of a construction site, and obtaining a building information model corresponding to the construction site; The embodiment of the present invention obtains on-site point cloud data of the construction site through a camera (or equipment such as a drone or robot that can be equipped with a camera), and obtains the building information model corresponding to the construction site from a database (such as a BIM collaborative management platform).

[0029] S202, calculating the average distance between each point cloud and the neighboring points in the on-site point cloud data using a K-nearest neighbor algorithm; The embodiment of the present invention calculates the relationship between each point cloud and its neighborhood by using the K-nearest neighbor algorithm. The average distance between points , ,in, For the point clouds, yes No. Neighborhood points, is the total number of neighborhood points. The nearest neighbor algorithm is an existing algorithm and will not be described in detail here.

[0030] S203, using Gaussian distribution and a preset noise reduction threshold calculation formula, based on the average distance, to calculate a noise reduction threshold; In an optional embodiment, the noise reduction threshold is calculated based on the average distance using a Gaussian distribution and a preset noise reduction threshold calculation formula, including: Using Gaussian distribution, calculate the mean and standard deviation of the average distance; The mean and the standard deviation are input into a preset noise reduction threshold calculation formula to calculate the noise reduction threshold.

[0031] The embodiment of the present invention assumes that the average distance obtained by the above calculation is Gaussian distribution. Through Gaussian distribution, the calculation formula corresponding to Gaussian distribution is as well as , then the corresponding mean can be obtained and standard deviation ,in, is the total number of point clouds in the point cloud data, the mean Reflects the overall density of the point cloud, standard deviation A measure of the dispersion of the average distance.

[0032] The mean and standard deviation are input into the preset noise reduction threshold calculation formula to calculate the noise reduction threshold. The preset noise threshold calculation formula is specifically: ; in, is the noise reduction threshold, is the proportional coefficient.

[0033] S204, removing point clouds with an average distance greater than the noise reduction threshold from the on-site point cloud data to obtain noise-reduced point cloud data; In an embodiment of the present invention, the average distance calculated above is compared with the noise reduction threshold, and point clouds with an average distance greater than the noise reduction threshold are removed from the point cloud data as outliers, thereby performing noise reduction on the point cloud data to obtain noise-reduced point cloud data.

[0034] S205, discretizing the building information model to obtain discrete point cloud data corresponding to the building information model; To align a BIM model (Building Information Model) with point cloud data from on-site scans, the present invention discretizes the BIM model. Specifically, the BIM model is converted into discrete point cloud data based on a preset initial average point spacing. The preset initial average point spacing can be set based on actual needs.

[0035] S206, resampling the discrete point cloud data according to the noise-reduced point cloud data to obtain resampled discrete point cloud data; In an optional embodiment, resampling the discrete point cloud data according to the noise-reduced point cloud data to obtain resampled discrete point cloud data includes: Calculating the average spacing of each point cloud in the noise-reduced point cloud data; The discrete point cloud data is resampled using the average spacing to obtain resampled discrete point cloud data.

[0036] The embodiment of the present invention calculates the average spacing of each point cloud in the noise-reduced point cloud data, and resamples the discrete point cloud data according to the average spacing of the point clouds in the noise-reduced point cloud data to match the density of the two point cloud data.

[0037] S207, using a density-adaptive voxel grid downsampling method to downsample the resampled discrete point cloud data and the de-noised point cloud data to obtain downsampled de-noised point cloud data and downsampled model point cloud data; In an optional embodiment, a density-adaptive voxel grid downsampling method is used to downsample the resampled discrete point cloud data and the de-noised point cloud data to obtain downsampled de-noised point cloud data and downsampled model point cloud data, including: Based on a preset voxel side length, dividing the resampled discrete point cloud data and the de-noised point cloud data into a plurality of three-dimensional voxel grids, respectively, to obtain a plurality of first three-dimensional voxel grid data corresponding to the resampled discrete point cloud data and a plurality of second three-dimensional voxel grid data corresponding to the de-noised point cloud data; respectively determining whether the number of point clouds in each of the first three-dimensional voxel grid data and each of the second three-dimensional voxel grid data is greater than or equal to a preset minimum number of voxel points; if so, determining the point cloud closest to the center point in the first three-dimensional voxel grid data or the second three-dimensional voxel grid data whose number of point clouds is greater than or equal to the preset minimum number of voxel points as a representative point cloud; if not, adjusting the voxel side length until the number of point clouds in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is greater than or equal to the preset minimum number of voxel points, and determining the point cloud closest to the center point in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data as the representative point cloud; Point clouds other than the representative point cloud are eliminated from the resampled discrete point cloud data and the denoised point cloud data to obtain downsampled denoised point cloud data and downsampled model point cloud data.

[0038] In an embodiment of the present invention, the discrete point cloud data after resampling still has the problem of a huge number of point clouds. In order to reduce the problem of registration difficulties caused by the huge number of point clouds, downsampling is performed using a density-adaptive voxel grid downsampling method to retain the key geometric features of the two point cloud data (such as key structural features such as beams, columns, and pipes).

[0039] Preset voxel edge length and the minimum number of voxel points , with voxel side length As a benchmark, the resampled discrete point cloud data and the noise-reduced point cloud data are divided into multiple three-dimensional voxel grids (the three-dimensional voxel grid is a grid with a side length of ), obtain a plurality of first three-dimensional voxel grid data corresponding to the resampled discrete point cloud data and a plurality of second three-dimensional voxel grid data corresponding to the denoised point cloud data; respectively determine whether the number of point clouds in each first three-dimensional voxel grid data and each second three-dimensional voxel grid data is greater than or equal to the minimum number of voxel points. ; When the number of point clouds in the first 3D voxel grid data or the second 3D voxel grid data is greater than or equal to the minimum number of voxel points When the number of point clouds in the first three-dimensional voxel grid data or the second three-dimensional voxel grid data is less than the minimum number of voxel points, the point cloud closest to the center point in the corresponding first three-dimensional voxel grid data or the second three-dimensional voxel grid data is determined as the representative point cloud; When , the voxel side length needs to be adjusted , and then obtain the number of point clouds in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data, so that the number of point clouds in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is greater than or equal to the minimum number of voxel points Then, a point cloud closest to the center point in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is determined as a representative point cloud.

[0040] The point clouds other than the representative point cloud are eliminated from the resampled discrete point cloud data and the denoised point cloud data, so that the cloud points in each three-dimensional voxel grid (including the first three-dimensional voxel grid data and the second three-dimensional voxel grid data) are reduced to one, that is, only the representative point cloud in each three-dimensional voxel grid is retained for downsampling, and the downsampled denoised point cloud data and the downsampled model point cloud data are obtained, thereby retaining the key geometric features in the two point cloud data.

[0041] S208, comparing the downsampled denoised point cloud data with actual site information of the construction site, and repairing missing data in the downsampled denoised point cloud data using a depth completion algorithm to obtain repaired denoised point cloud data; In an embodiment of the present invention, due to the complex construction site environment and factors such as occlusion of certain parts of buildings, some point cloud data may not be captured and may be missing. This missing point cloud may affect the final registration result, thereby affecting the construction quality acceptance result information. Therefore, the downsampled denoised point cloud data is compared with the actual site information of the construction site to determine the missing point cloud data (missing data) in the downsampled denoised point cloud data. A depth completion algorithm is used to calculate the depth of missing point A by projecting it into the world coordinate system and reverse mapping it to the target plane. The depth is then compared for consistency with the depth of reference point B (reference point B is a valid point cloud adjacent to missing point A on the same target plane. The point cloud can be obtained by searching the same target plane in the downsampled denoised point cloud data using the K-nearest neighbor algorithm). If the relative error is less than a preset repair threshold (the preset repair threshold can be set according to actual needs), the depth estimate is determined to be valid. The depth value is then used to determine the corresponding plane and complete the missing point. Finally, the incomplete point cloud data is repaired through plane constraints and depth consistency, resulting in the repaired denoised point cloud data.

[0042] S209, performing coarse registration on the repaired denoised point cloud data and the downsampled model point cloud data by a feature point matching method, so as to align point clouds at corresponding positions in the repaired denoised point cloud data and the downsampled model point cloud data; The embodiment of the present invention performs coarse registration by a feature point matching method, and can align the point clouds at corresponding positions in the repaired denoised point cloud data and the downsampled model point cloud data.

[0043] 1) Endpoint detection: For each point cloud A in the repaired denoised point cloud data, calculate the projection of its neighboring point cloud on the tangent plane, and connect A with the neighboring point cloud to form an angle set. If the maximum angle , then A is determined to be an edge endpoint.

[0044] 2) Eigenvector construction: Select an endpoint A, find its closest point B and farthest point C, and construct three vectors: , , ; thus forming a feature vector set (the restored denoised point cloud data) and (Model point cloud data after downsampling).

[0045] 3) Orthogonal alignment: through orthogonal matrix and Align the eigenvectors: , , .

[0046] Calculate the rotation matrix and translation matrices ,in, is the edge endpoint of the repaired denoised point cloud data, It is the edge endpoint of the downsampled model point cloud data.

[0047] 4) Application and The downsampled model point cloud data is preliminarily aligned to the repaired denoised point cloud data to reduce the number of subsequent ICP algorithm iterations.

[0048] S210, using the FREAK algorithm and the ICP algorithm, performing precise registration on the aligned repaired denoised point cloud data and the downsampled model point cloud data to obtain construction quality acceptance result information of the construction site; The embodiment of the present invention adopts ICP combined with FREAK to perform point cloud registration. FREAK is a binary descriptor that can be used to represent feature points in an image or point cloud, and can improve the accuracy of feature point matching and registration efficiency. Using the FREAK algorithm, a retina-like circular sampling grid is established around each key point for the repaired denoised point cloud data and the downsampled model point cloud data. The sampling points in the center area of ​​the circular sampling grid are dense, and the sampling points in the edge area are sparse. Predefined sampling point pairs are selected from the circular sampling grid, and the grayscale difference between each pair of points is calculated. If the grayscale value of the first point is greater than that of the second point, the corresponding bit in the descriptor is set to 1, otherwise it is set to 0. By performing this binary comparison on a large number of sampling point pairs, a compact binary descriptor is finally generated. The binary descriptors of the two point clouds are matched using the Hamming distance. The feature points of the repaired denoised point cloud data and the downsampled model point cloud data after coarse registration are extracted using the same binary descriptor. After filtering, the corresponding rotation matrix R and translation matrix T are constructed according to the selected direction to perform coordinate transformation of the two point clouds (the repaired denoised point cloud data and the downsampled model point cloud data). The average distance between the corresponding positions of the two point clouds after coordinate transformation is calculated, and this step is iterated continuously. When the average distance between the two point cloud data is less than the preset registration distance threshold, or the number of iterations is greater than the preset registration threshold, the final registration is completed, and the registration information of the construction site and the corresponding BIM model is obtained. The distance between the corresponding point clouds at the same construction location is calculated from the registration information, thereby calculating the deviation of each key structure (such as beams, columns, pipes, etc.) in the two point cloud data. The deviation of each key structure (the distance between the corresponding point clouds at the same construction location) is used to generate the construction quality acceptance result information of the construction site. If the distance between the corresponding point clouds at the same construction location is less than or equal to the preset acceptance distance threshold, or the average deviation of each cloud point in the same key structure (the average distance between each point cloud in the same structure) is less than or equal to the preset acceptance deviation threshold, the construction quality is determined to be qualified. Otherwise, the construction quality of the corresponding structure is determined to be unqualified. Therefore, the construction quality acceptance results of each structure are recorded to generate the construction quality acceptance result information of the construction site.

[0049] In the construction quality acceptance result information, if the point cloud distance or construction deviation is too large (greater than the preset acceptance distance threshold or the preset acceptance deviation threshold), the corresponding position in the coordinate system is mapped to the BIM model for differentiated display, and prompts and feedback are provided on the BIM model, so as to achieve the purpose of remote inspection and acceptance by management personnel.

[0050] A construction quality acceptance method disclosed in an embodiment of the present invention utilizes a building information model, a FREAK algorithm, and an ICP algorithm to inspect a construction site, thereby improving the efficiency of project acceptance and laying an important foundation for accurate acceptance of construction quality. Example 3

[0051] See also Figure 3 , Figure 3 This is a structural block diagram of an embodiment of a construction quality acceptance device of the present invention, the device comprising: An acquisition module 301 is used to acquire on-site point cloud data of a construction site and a building information model corresponding to the construction site; A noise reduction module 302 is used to perform noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data; A downsampling module 303 is configured to perform downsampling processing on the de-noised point cloud data and the building information model to obtain downsampled de-noised point cloud data and downsampled model point cloud data; The registration module 304 is used to perform point cloud registration on the downsampled denoised point cloud data and the downsampled model point cloud data using the FREAK algorithm and the ICP algorithm to obtain the construction quality acceptance result information of the construction site.

[0052] In an optional embodiment, the noise reduction module 302 includes: The first calculation submodule is used to calculate the average distance between each point cloud and the neighboring points in the on-site point cloud data by using a K-nearest neighbor algorithm; A second calculation submodule is configured to calculate a noise reduction threshold based on the average distance using a Gaussian distribution and a preset noise reduction threshold calculation formula; The elimination submodule is used to eliminate the point cloud whose average distance is greater than the noise reduction threshold from the on-site point cloud data to obtain noise-reduced point cloud data.

[0053] In an optional embodiment, the second calculation submodule includes: A first calculation unit is used to calculate the mean and standard deviation of the average distance using Gaussian distribution; The second calculation unit is used to input the mean and the standard deviation into a preset noise reduction threshold calculation formula to calculate the noise reduction threshold In an optional embodiment, the downsampling module 303 includes: a discretization submodule, configured to discretize the building information model to obtain discrete point cloud data corresponding to the building information model; a resampling submodule, configured to resample the discrete point cloud data according to the noise reduction point cloud data to obtain resampled discrete point cloud data; The downsampling submodule is used to downsample the resampled discrete point cloud data and the de-noised point cloud data using a density-adaptive voxel grid downsampling method to obtain downsampled de-noised point cloud data and downsampled model point cloud data.

[0054] In an optional embodiment, the resampling submodule includes: A third calculation unit is used to calculate the average spacing of each point cloud in the noise reduction point cloud data; The resampling unit is configured to resample the discrete point cloud data using the average spacing to obtain resampled discrete point cloud data.

[0055] In an optional embodiment, the downsampling submodule includes: a dividing unit, configured to divide the resampled discrete point cloud data and the de-noised point cloud data into a plurality of three-dimensional voxel grids based on a preset voxel side length, to obtain a plurality of first three-dimensional voxel grid data corresponding to the resampled discrete point cloud data and a plurality of second three-dimensional voxel grid data corresponding to the de-noised point cloud data; a judging unit, configured to respectively judge whether the number of point clouds in each of the first three-dimensional voxel grid data and each of the second three-dimensional voxel grid data is greater than or equal to a preset minimum number of voxel points; if so, determining a point cloud closest to the center point in the first three-dimensional voxel grid data or the second three-dimensional voxel grid data whose number of point clouds is greater than or equal to the preset minimum number of voxel points as a representative point cloud; if not, adjusting a voxel side length until the number of point clouds in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is greater than or equal to the preset minimum number of voxel points, and determining a point cloud closest to the center point in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data as the representative point cloud; A culling unit is used to cull the point clouds other than the representative point cloud from the resampled discrete point cloud data and the de-noised point cloud data, to obtain the downsampled de-noised point cloud data and the downsampled model point cloud data. In an optional embodiment, the registration module 304 includes: a comparison submodule, configured to compare the downsampled denoised point cloud data with the actual site information of the construction site, and repair missing data in the downsampled denoised point cloud data using a depth completion algorithm to obtain repaired denoised point cloud data; a coarse registration submodule, configured to perform coarse registration on the repaired denoised point cloud data and the downsampled model point cloud data by a feature point matching method, so as to align point clouds at corresponding positions in the repaired denoised point cloud data and the downsampled model point cloud data; The fine registration submodule is used to use the FREAK algorithm and the ICP algorithm to finely register the aligned repaired denoised point cloud data and the downsampled model point cloud data to obtain the construction quality acceptance result information of the construction site. Example 4

[0056] An embodiment of the present invention also provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of a construction quality acceptance method of any embodiment. Example 5

[0057] An embodiment of the present invention further provides a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by the processor, the steps of a construction quality acceptance method of any embodiment are implemented.

[0058] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0059] In the several embodiments provided in this application, it should be understood that the methods, devices, electronic devices and storage media disclosed in the present invention can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0060] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0061] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0062] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned readable storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0063] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A construction quality acceptance method, characterized in that: include: Acquiring on-site point cloud data of a construction site and acquiring a building information model corresponding to the construction site; Performing noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data; Downsampling the noise-reduced point cloud data and the building information model to obtain downsampled noise-reduced point cloud data and downsampled model point cloud data; The downsampled denoised point cloud data and the downsampled model point cloud data are aligned using the FREAK algorithm and the ICP algorithm to obtain construction quality acceptance result information of the construction site.

2. The construction quality acceptance method according to claim 1, characterized in that: Performing noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data, including: The average distance between each point cloud and the neighboring points in the on-site point cloud data is calculated by the K-nearest neighbor algorithm; Calculating a noise reduction threshold based on the average distance using a Gaussian distribution and a preset noise reduction threshold calculation formula; Point clouds with an average distance greater than the noise reduction threshold are removed from the on-site point cloud data to obtain noise-reduced point cloud data.

3. The construction quality acceptance method according to claim 2, characterized in that: The noise reduction threshold is calculated based on the average distance using Gaussian distribution and a preset noise reduction threshold calculation formula, including: Using Gaussian distribution, calculate the mean and standard deviation of the average distance; The mean and the standard deviation are input into a preset noise reduction threshold calculation formula to calculate the noise reduction threshold.

4. The construction quality acceptance method according to claim 1, characterized in that: Downsampling the noise-reduced point cloud data and the building information model to obtain downsampled noise-reduced point cloud data and downsampled model point cloud data, including: Discretizing the building information model to obtain discrete point cloud data corresponding to the building information model; resampling the discrete point cloud data according to the noise reduction point cloud data to obtain resampled discrete point cloud data; A density-adaptive voxel grid downsampling method is used to downsample the resampled discrete point cloud data and the denoised point cloud data to obtain downsampled denoised point cloud data and downsampled model point cloud data.

5. The construction quality acceptance method according to claim 4, characterized in that: Resampling the discrete point cloud data according to the noise reduction point cloud data to obtain resampled discrete point cloud data includes: Calculating the average spacing of each point cloud in the noise-reduced point cloud data; The discrete point cloud data is resampled using the average spacing to obtain resampled discrete point cloud data.

6. The construction quality acceptance method according to claim 4, characterized in that: A density-adaptive voxel grid downsampling method is used to downsample the resampled discrete point cloud data and the de-noised point cloud data to obtain downsampled de-noised point cloud data and downsampled model point cloud data, including: Based on a preset voxel side length, dividing the resampled discrete point cloud data and the de-noised point cloud data into a plurality of three-dimensional voxel grids, respectively, to obtain a plurality of first three-dimensional voxel grid data corresponding to the resampled discrete point cloud data and a plurality of second three-dimensional voxel grid data corresponding to the de-noised point cloud data; respectively determining whether the number of point clouds in each of the first three-dimensional voxel grid data and each of the second three-dimensional voxel grid data is greater than or equal to a preset minimum number of voxel points; if so, determining the point cloud closest to the center point in the first three-dimensional voxel grid data or the second three-dimensional voxel grid data whose number of point clouds is greater than or equal to the preset minimum number of voxel points as a representative point cloud; if not, adjusting the voxel side length until the number of point clouds in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data is greater than or equal to the preset minimum number of voxel points, and determining the point cloud closest to the center point in the adjusted first three-dimensional voxel grid data or the adjusted second three-dimensional voxel grid data as the representative point cloud; Point clouds other than the representative point cloud are eliminated from the resampled discrete point cloud data and the denoised point cloud data to obtain downsampled denoised point cloud data and downsampled model point cloud data.

7. The construction quality acceptance method according to claim 1, characterized in that: Performing point cloud registration on the downsampled denoised point cloud data and the downsampled model point cloud data using the FREAK algorithm and the ICP algorithm to obtain construction quality acceptance result information of the construction site, including: Comparing the downsampled denoised point cloud data with actual site information of the construction site, and repairing missing data in the downsampled denoised point cloud data using a depth completion algorithm to obtain repaired denoised point cloud data; performing coarse registration on the repaired denoised point cloud data and the downsampled model point cloud data by a feature point matching method, so as to align point clouds at corresponding positions in the repaired denoised point cloud data and the downsampled model point cloud data; The FREAK algorithm and the ICP algorithm are used to perform precise registration on the aligned repaired denoised point cloud data and the downsampled model point cloud data to obtain construction quality acceptance result information of the construction site.

8. A construction quality acceptance device, characterized in that: include: An acquisition module, configured to acquire on-site point cloud data of a construction site and a building information model corresponding to the construction site; A noise reduction module, configured to perform noise reduction processing on the on-site point cloud data to obtain noise-reduced point cloud data; a downsampling module, configured to perform downsampling processing on the de-noised point cloud data and the building information model to obtain downsampled de-noised point cloud data and downsampled model point cloud data; The registration module is used to perform point cloud registration on the downsampled denoised point cloud data and the downsampled model point cloud data through the FREAK algorithm and the ICP algorithm to obtain the construction quality acceptance result information of the construction site.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 7 is executed.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is executed.

Images