A method, device and equipment for registering BIM model and engineering point cloud

Abstract

The present application discloses a registration method, device and equipment for a BIM model and an engineering point cloud. After a first target point cloud collected based on an engineering scene and a source point cloud converted from a BIM model, the first target point cloud and the source point cloud are roughly registered based on a pre-selected edge endpoint to obtain a second target point cloud. Then the second target point cloud and the source point cloud are accurately registered to obtain a precisely registered point cloud. The initial BIM model is superimposed on the precisely registered point cloud to obtain a target BIM model, and the target BIM model is spatially aligned with the engineering scene. The present application uses the feature correspondence and spatial relationship between point clouds for registration, reduces the feature point recognition time, and improves the efficiency of point cloud registration. The edge endpoints are used as feature points, the point cloud point information is fully considered, the accuracy of the point cloud edge fitting is improved, and the accuracy of the point cloud registration is improved; based on the point cloud registration transformation parameters, the BIM model is aligned with the real space scene to achieve the virtual-real fusion of the BIM model and the real scene.

Description

Technical Field

[0001] The present application relates to the technical field of engineering construction management, and more specifically, to a method, device and equipment for aligning a BIM model with an engineering point cloud. Background Art

[0002] Engineering construction inspection is an important means to maintain engineering safety and high-quality construction. At present, the inspection information in engineering construction quality inspection is not intuitive, and some working surfaces and construction units are difficult to be clearly visible, making it difficult for managers to make accurate quality judgments. Therefore, enhancing the visualization of inspection units and simplifying the retrieval method of demand information are of great practical value to improve the quality and efficiency of inspection.

[0003] AR technology can integrate virtual information into real scenes. Through AR technology, BIM models can be superimposed and interacted with real engineering scenes in real time, which can effectively enhance the inspection personnel's understanding and perception of actual projects and improve the quality of inspections. The registration accuracy of BIM models and real engineering scenes is one of the important reasons that limit the implementation of AR quality inspections. In addition, the complex environment of engineering construction scenes will affect the accuracy and efficiency of the registration of BIM models and real scenes.

[0004] At present, when the BIM model is aligned with the real engineering scene, a large number of feature points need to be identified and matched. Therefore, it takes a lot of time and is inefficient to align a large amount of BIM model point cloud data with the real engineering scene point cloud data. Moreover, when the BIM model point cloud has rich edge features, the existing methods do not consider the point cloud point information, resulting in low alignment accuracy. Summary of the invention

[0005] In view of this, the present application provides a method, device, equipment and storage medium for registering a BIM model with an engineering point cloud, which is used to solve the problem that a large number of feature points need to be identified and matched when registering an existing BIM model with an engineering point cloud, resulting in low registration efficiency. In addition, the existing registration method does not consider the point cloud point information, resulting in low registration accuracy.

[0006] To achieve the above objectives, the proposed solution is as follows:

[0007] A registration method for a BIM model and an engineering point cloud, comprising:

[0008] Acquire point cloud data of the engineering scene to obtain the first target point cloud;

[0009] Convert the initial BIM model corresponding to the engineering scene into point cloud data to obtain the source point cloud;

[0010] Roughly aligning the first target point cloud with the source point cloud based on the pre-selected edge endpoints to obtain a second target point cloud;

[0011] Accurately align the second target point cloud with the source point cloud to obtain a precisely aligned point cloud;

[0012] The initial BIM model is superimposed on the precisely registered point cloud to obtain the target BIM model;

[0013] Spatial alignment of the target BIM model with the project scene.

[0014] Preferably, roughly registering the first target point cloud with the source point cloud based on the pre-selected edge endpoints to obtain the second target point cloud includes:

[0015] Determine the closest feature point and the farthest feature point in the two point clouds based on the pre-selected edge endpoints in the first target point cloud and the source point cloud respectively;

[0016] Based on the edge endpoints, the nearest feature points and the farthest feature points in the point cloud, feature vector sets of the first target point cloud and the source point cloud are respectively constructed to obtain a target feature vector set and a source feature vector set;

[0017] The target feature vector set and the source feature vector set are respectively transformed so that the transformed target feature vector set coincides with the source vector set, thereby obtaining a first orthogonal matrix and a second orthogonal matrix;

[0018] Determine an initial point cloud transformation matrix based on the first orthogonal matrix and the second orthogonal matrix;

[0019] The first target point cloud is transformed according to the initial point cloud transformation matrix to obtain a second target point cloud.

[0020] Preferably, the second target point cloud and the source point cloud are accurately registered to obtain a precisely registered point cloud, including:

[0021] Obtain candidate feature point sets of the second target point cloud and the source point cloud respectively;

[0022] The feature points in the candidate feature point sets of the second target point cloud and the source point cloud are screened respectively to obtain initial feature point sets of the two point clouds;

[0023] The optimal transformation matrix is ​​obtained by calculating the initial feature point set of the second target point cloud and the source point cloud;

[0024] Transform the feature points in the initial feature point set of the second target point cloud according to the optimal transformation matrix to obtain a target feature point set;

[0025] Constructing a third target point cloud based on the target feature point set;

[0026] Calculate the average distance between the third target point cloud and the source point cloud;

[0027] Determine whether the average distance between the third target point cloud and the source point cloud is less than a preset threshold;

[0028] If yes, the third target point cloud is output as the precisely registered point cloud.

[0029] Preferably, the feature points in the candidate feature point sets of the second target point cloud and the source point cloud are screened respectively, including:

[0030] Construct candidate feature point matrices for the second target point cloud and the source point cloud respectively, where each row in the candidate feature point matrix represents a feature point, and the pixels of the feature points are used as elements of the matrix;

[0031] Calculate the average value of each column in each candidate feature point matrix;

[0032] Calculate the difference between the average value of each column of each candidate feature point matrix and the preset standard value;

[0033] The first N columns with the smallest absolute value of the difference are taken as the feature points in the initial feature point set, where N is a positive integer greater than 0, and N is a positive integer greater than 0.

[0034] Preferably, after determining whether the average distance between the third target point cloud and the source point cloud is less than a preset threshold, the method further includes:

[0035] If not, the process returns to the step of respectively acquiring candidate feature point sets of the second target point cloud and the source point cloud until the average distance between the third target point cloud and the source point cloud is less than a preset threshold.

[0036] Preferably, converting the initial BIM model corresponding to the engineering scene into point cloud data includes:

[0037] Store the initial BIM model in OBJ format;

[0038] The initial BIM model in OBJ format is converted into PCD format to obtain the source point cloud.

[0039] A BIM model and engineering point cloud registration device, comprising:

[0040] An engineering point cloud acquisition unit, used to acquire point cloud data of an engineering scene to obtain a first target point cloud;

[0041] A source point cloud acquisition unit is used to convert the initial BIM model corresponding to the engineering scene into point cloud data to obtain a source point cloud;

[0042] A coarse registration unit, used for roughly registering the first target point cloud with the source point cloud based on pre-selected edge endpoints to obtain a second target point cloud;

[0043] A precise registration unit, used for accurately registering the second target point cloud with the source point cloud to obtain a precise registration point cloud;

[0044] A point cloud overlay unit is used to overlay the initial BIM model onto the precisely registered point cloud to obtain a target BIM model;

[0045] The model alignment unit is used to spatially align the target BIM model with the engineering scene.

[0046] A BIM model and engineering point cloud registration device, comprising: a memory and a processor;

[0047] The memory is used to store programs;

[0048] The processor is used to execute the program to implement the various steps of the aforementioned BIM model and engineering point cloud registration method.

[0049] It can be seen from the above technical solutions that the registration method of the BIM model and the engineering point cloud provided in this application, after obtaining the first target point cloud collected based on the engineering scene and the source point cloud converted from the BIM model, first roughly aligns the first target point cloud and the source point cloud based on the pre-selected edge endpoints to obtain the second target point cloud. Then the second target point cloud and the source point cloud are accurately registered to obtain a precisely registered point cloud. Finally, the initial BIM model is superimposed on the precisely registered point cloud to obtain the target BIM model, and the target BIM model is spatially aligned with the engineering scene. This application utilizes the point cloud of the constructed BIM model and the point cloud of the real engineering scene, and aligns the point cloud according to the feature correspondence and spatial relationship between the point cloud and the point cloud, which reduces the feature point recognition time and improves the efficiency of point cloud registration.

[0050] This application uses the edge endpoints of the point cloud as feature points, fully considers the point cloud point information, improves the accuracy of point cloud edge fitting, and improves the accuracy of point cloud registration. Based on the point cloud registration transformation parameters, this application accurately aligns the BIM model with the real space scene. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0052] Figure 1 A flowchart of a method for aligning a BIM model with an engineering point cloud provided in an embodiment of the present application;

[0053] Figure 2a-2c A point cloud and model schematic diagram of the laboratory building provided in the embodiment of the present application;

[0054] Figure 3 A rough point cloud registration flow chart provided in an embodiment of the present application;

[0055] Figure 4 A point cloud precise registration flow chart provided in an embodiment of the present application;

[0056] Figure 5a-Figure 5b A schematic diagram of point cloud overlay and BIM model alignment provided in an embodiment of the present application;

[0057] Figure 6 A schematic diagram of a BIM model and engineering point cloud registration device provided in an embodiment of the present application;

[0058] Figure 7 A hardware structure block diagram of a BIM model and engineering point cloud registration device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0060] First, combine Figure 1 A method for aligning a BIM model with an engineering point cloud provided in an embodiment of the present application is introduced. Figure 1 As shown, the method may include:

[0061] Step S01, obtaining point cloud data of the engineering scene.

[0062] Specifically, point cloud data of a real engineering scene is collected through AR mobile display devices such as AR glasses, mobile phones, and tablet computers, and the collected point cloud data of the engineering scene is stored to obtain the first target point cloud. For example, taking a laboratory building under construction in a certain project as an example, point cloud data of the laboratory building is collected to obtain the following: Figure 2a The first target point cloud is shown.

[0063] Step S02: converting the initial BIM model corresponding to the engineering scene into point cloud data.

[0064] Specifically, a BIM model of the engineering scene is constructed according to the design file, and the constructed BIM model is used as the initial BIM model. The initial BIM model includes detailed information such as the size and material of the project. The initial BIM model is stored in OBJ format; then the initial BIM model in OBJ format is converted into the PCD format of point cloud data to obtain the source point cloud. For example: According to the design file of the laboratory building, a BIM model of the laboratory building is constructed to obtain the initial BIM model of the laboratory building. Figure 2b Then, the initial BIM model is converted into point cloud data through CloudCompare software, and the source point cloud of the experimental building is obtained as shown in Figure 2c shown.

[0065] Step S03: roughly aligning the first target point cloud and the source point cloud based on the pre-selected edge endpoints.

[0066] Specifically, Figure 3 As shown, the endpoints of the edges are selected in the first target point cloud and the source point cloud, and the edge endpoints are determined as feature points.

[0067] Based on the pre-selected edge endpoints in the first target point cloud and the source point cloud, the nearest feature point and the farthest feature point in the two point clouds are determined respectively. The distances of each point in the two point clouds to the pre-selected edge endpoints are sorted respectively. In order to eliminate the influence of abnormal points on the nearest feature point and the farthest feature point of the edge point, the average of the nearest and farthest l points after sorting is selected as the nearest feature point and the farthest feature point. That is: l = ceiling (M, 1), where ceiling () is an upward rounding function, and M represents the total number of points in the point cloud.

[0068] Based on the edge endpoints, the nearest feature points and the farthest feature points in the point cloud, the feature vector sets of the first target point cloud and the source point cloud are respectively constructed to obtain the target feature vector set and the source feature vector set. p and the source feature vector set n of the source point cloud q .in, The first eigenvector n1 and the second eigenvector are the eigenvectors of the first target point cloud and the source point, respectively, which take the cloud edge endpoint as the starting point and point to the nearest feature point and the farthest feature point in the corresponding point cloud. The third eigenvector n3 is the vector sum of the first eigenvector n1 and the second eigenvector n2, n3=n1+n2.

[0069] The target feature vector set and the source feature vector set are transformed respectively so that the transformed target feature vector set coincides with the source vector set, and a first orthogonal matrix and a second orthogonal matrix are obtained. The target feature vector set is roughly coincident with the source feature vector set through a certain transformation, that is, the target feature vector set and the source feature vector set satisfy n p =R 0 nq . It can be obtained by the first orthogonal matrix W p and the second orthogonal matrix W q To convert, that is, through For the target feature vector set n p and the source feature vector set n of the source point cloud q If the target feature vector set after transformation coincides with the source feature vector set, then n q ′=n p ′, and satisfies W p T W p =W p W p T =I;W q ∈R 3×3 , W q T W q =W q W q T =I. The initial point cloud transformation matrix is ​​determined based on the first orthogonal matrix and the second orthogonal matrix. The initial point cloud transformation matrix includes a first rotation matrix and a first translation matrix. 0 =W p -1 W q , according to the first orthogonal matrix W p and the second orthogonal matrix W q , find the first rotation matrix R of the transformation 0 The first translation matrix can be obtained by converting the coordinates of the edge endpoints of the point cloud, that is, T 0 =AR 0 A', where A and A' are the edge endpoints of the first target point cloud and the source point cloud respectively, and the first translation matrix T is obtained by the coordinate value. 0 .

[0070] The first target point cloud is transformed according to the initial point cloud transformation matrix to obtain the second target point cloud. 0 and the first translation matrix T 0 Transformation is performed to obtain the second target point cloud after rough registration.

[0071] Step S04: accurately align the second target point cloud with the source point cloud.

[0072] Specifically, the second target point cloud is used as the target point cloud, and candidate feature point sets of the second target point cloud and the source point cloud are obtained respectively. Figure 4 As shown, the candidate feature point set p in the second target point cloud P is extracted respectively. i and the candidate feature point set q in the source point cloud Qi .

[0073] The feature points in the candidate feature point sets of the second target point cloud and the source point cloud are screened respectively to obtain the initial feature point sets of the two point clouds. Binary descriptors of feature points are extracted from the target point cloud and the source point cloud, and the feature points are screened after extraction. The screening process can be:

[0074] ① Construct candidate feature point matrices for the second target point cloud and the source point cloud respectively. Each row in the candidate feature point matrix represents a feature point, and the pixels of the feature points are used as elements of the matrix.

[0075] ② Calculate the average value of each column in each candidate feature point matrix.

[0076] ③ Calculate the difference between the average value of each column of each candidate feature point matrix and the preset standard value.

[0077] ④ Take the first N columns with the smallest absolute value of the difference as the feature points in the initial feature point set, where N is a positive integer greater than 0. For example, when the average value is close to 0.5, it means that the variance is large and the feature points carry rich information. Therefore, set the standard value to 0.5, calculate the difference between the average value of each column in the matrix and 0.5, and arrange the columns in the matrix from small to large according to the absolute value of the difference, and select the feature points of the first N columns in the matrix as the feature points in the initial feature point set.

[0078] The optimal transformation matrix is ​​obtained by calculating the initial feature point set of the second target point cloud and the source point cloud. After the feature points are screened, the second target point cloud and the source point cloud are unified to the same reference coordinate system through rotation and translation transformation. The second target point cloud can be rotated and translated through the second rotation matrix and the second translation matrix. The optimal transformation matrix can be obtained by solving, and the optimal transformation matrix includes the second rotation matrix R and the second translation matrix T. According to the candidate feature point set p i And the candidate feature point set q i Construct the optimal transformation function E(R, T), where E(R, T) represents the error between the target point cloud p and the source point cloud set after transformation by the transformation matrix. Then, the optimal transformation matrix is ​​obtained by finding the optimal solution (R, T) that satisfies min(E(R, T)).

[0079] Optimal conversion function:

[0080] The second translation matrix: T = [t x t y t z ] T .

[0081] Second rotation matrix:

[0082] where α, β, and γ respectively represent the rotation angles of the feature points along the x, y, and z axes, and t x , t y , t z represent the translation amounts of the feature points along the x, y, and z axes respectively.

[0083] Transform the feature points in the initial feature point set of the second target point cloud according to the optimal transformation matrix to obtain the target feature point set. Transform the feature points in the initial feature point set according to the second rotation matrix R and the second translation matrix T to obtain the target feature point set.

[0084] Construct a third target point cloud based on the target feature point set. Then calculate the average distance between the third target point cloud and the source point cloud. After transforming the initial feature point set of the second target point cloud according to the optimal transformation matrix, obtain the target feature point set. After constructing the third target point cloud according to the target feature point set, calculate the average distance d between the third target point cloud and the source point cloud, where in the formula, m represents the number of points in the third target point cloud, q′ i represents a point in the third target point cloud, and p i represents a point in the source point cloud.

[0085] Judge whether the average distance between the third target point cloud and the source point cloud is less than a preset threshold. Set the iteration threshold with the average distance d between the two point clouds as the constraint condition, and set the average distance accuracy between the two point clouds as the constraint condition to be e, that is: when d < e, the iteration ends. At this time, the point cloud precise registration is completed, and the third target point cloud is output as the precisely registered point cloud. If d ≥ e, then return to execute the steps of respectively obtaining the candidate feature point sets of the second target point cloud and the source point cloud until the average distance between the third target point cloud and the source point cloud is less than the preset threshold e. Example: Output the precisely registered point cloud of the experimental building as Figure 5a shown.

[0086] Step S05, superimpose the initial BIM model on the precisely registered point cloud.

[0087] Specifically, unify the point cloud data of the precisely registered point cloud and the model data of the BIM model into the same coordinate system, add the three-dimensional coordinate information of the BIM model to the point cloud data, and superimpose the BIM model with the registered precisely registered point cloud to obtain the target BIM model.

[0088] Step S06, spatially align the target BIM model with the engineering scene.

[0089] Specifically, the target BIM model with superimposed precise registration point cloud is integrated with the real engineering scene. That is, the transformation matrix of point cloud registration is used to design the AR interactive scene, and the target BIM model is spatially aligned with the engineering scene through the position and posture information provided by the mobile device. This realizes the BIM model augmented reality visualization inspection. For example, the target BIM model of the laboratory building is spatially aligned with the engineering scene. Figure 5b shown.

[0090] The registration method of the BIM model and the engineering point cloud provided in the present application, after obtaining the first target point cloud collected based on the engineering scene and the source point cloud converted from the BIM model, first roughly aligns the first target point cloud and the source point cloud based on the pre-selected edge endpoints to obtain the second target point cloud. Then the second target point cloud and the source point cloud are accurately registered to obtain a precisely registered point cloud. Finally, the initial BIM model is superimposed on the precisely registered point cloud to obtain the target BIM model, and the target BIM model is spatially aligned with the engineering scene. The present application utilizes the point cloud of the constructed BIM model and the point cloud of the real engineering scene, and aligns the point cloud according to the feature correspondence and spatial relationship between the point cloud and the point cloud, thereby reducing the feature point recognition time and improving the efficiency of point cloud registration.

[0091] This application uses the edge endpoints of the point cloud as feature points, fully considers the point cloud point information, improves the accuracy of point cloud edge fitting, and improves the accuracy of point cloud registration. Based on the point cloud registration transformation parameters, this application accurately aligns the BIM model with the real space scene.

[0092] The following is a description of the BIM model and engineering point cloud registration device provided in an embodiment of the present application. The BIM model and engineering point cloud registration device described below and the BIM model and engineering point cloud registration method described above can be referenced to each other.

[0093] First combine Figure 6 , introduces the registration device between BIM model and engineering point cloud, such as Figure 6 As shown, the BIM model and the engineering point cloud registration device may include:

[0094] The engineering point cloud acquisition unit 100 is used to acquire point cloud data of the engineering scene to obtain a first target point cloud;

[0095] The source point cloud acquisition unit 200 is used to convert the initial BIM model corresponding to the engineering scene into point cloud data to obtain a source point cloud;

[0096] A coarse registration unit 300 is used to roughly register the first target point cloud and the source point cloud based on the pre-selected edge endpoints to obtain a second target point cloud;

[0097] A precise registration unit 400 is used to precisely register the second target point cloud with the source point cloud to obtain a precisely registered point cloud;

[0098] The point cloud overlay unit 500 is used to overlay the initial BIM model onto the precisely registered point cloud to obtain a target BIM model;

[0099] The model alignment unit 600 is used to spatially align the target BIM model with the engineering scene.

[0100] The BIM model and engineering point cloud registration device provided in the embodiment of the present application can be applied to the BIM model and engineering point cloud registration equipment. Figure 7 The hardware structure diagram of the BIM model and engineering point cloud registration equipment is shown in Figure 1. Figure 7 , the hardware structure of the device may include: at least one processor 1, at least one communication interface 2, at least one memory 3 and at least one communication bus 4;

[0101] In the embodiment of the present application, the number of the processor 1, the communication interface 2, the memory 3, and the communication bus 4 is at least one, and the processor 1, the communication interface 2, and the memory 3 communicate with each other through the communication bus 4;

[0102] The processor 1 may be a central processing unit CPU, or an application-specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention, etc.;

[0103] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory;

[0104] Among them, the memory stores a program, and the processor can call the program stored in the memory, and the program is used to implement various processing flows in the aforementioned BIM model and engineering point cloud registration solution.

[0105] Finally, it should be noted that, in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. In this specification, various embodiments are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the various embodiments can be referred to each other.

[0106] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A registration method for a BIM model and an engineering point cloud, characterized in that: include: Acquire point cloud data of the engineering scene to obtain the first target point cloud; Convert the initial BIM model corresponding to the engineering scene into point cloud data to obtain the source point cloud; Roughly aligning the first target point cloud with the source point cloud based on the pre-selected edge endpoints to obtain a second target point cloud; Accurately align the second target point cloud with the source point cloud to obtain a precisely aligned point cloud; The initial BIM model is superimposed on the precisely registered point cloud to obtain the target BIM model; Spatially align the target BIM model with the engineering scene; The first target point cloud and the source point cloud are roughly aligned based on the pre-selected edge endpoints to obtain the second target point cloud, including: Determine the closest feature point and the farthest feature point in the two point clouds based on the pre-selected edge endpoints in the first target point cloud and the source point cloud respectively; Based on the edge endpoints, the nearest feature points and the farthest feature points in the point cloud, feature vector sets of the first target point cloud and the source point cloud are respectively constructed to obtain a target feature vector set and a source feature vector set; The target feature vector set and the source feature vector set are respectively transformed so that the transformed target feature vector set coincides with the source vector set, thereby obtaining a first orthogonal matrix and a second orthogonal matrix; Determine an initial point cloud transformation matrix based on the first orthogonal matrix and the second orthogonal matrix; The first target point cloud is transformed according to the initial point cloud transformation matrix to obtain a second target point cloud.

2. The registration method of the BIM model and the engineering point cloud according to claim 1, characterized in that: The second target point cloud and the source point cloud are accurately registered to obtain a precisely registered point cloud, including: Obtain candidate feature point sets of the second target point cloud and the source point cloud respectively; The feature points in the candidate feature point sets of the second target point cloud and the source point cloud are screened respectively to obtain initial feature point sets of the two point clouds; The optimal transformation matrix is ​​obtained by calculating the initial feature point set of the second target point cloud and the source point cloud; Transform the feature points in the initial feature point set of the second target point cloud according to the optimal transformation matrix to obtain a target feature point set; Constructing a third target point cloud based on the target feature point set; Calculate the average distance between the third target point cloud and the source point cloud; Determine whether the average distance between the third target point cloud and the source point cloud is less than a preset threshold; If yes, the third target point cloud is output as the precisely registered point cloud.

3. The registration method of a BIM model and an engineering point cloud according to claim 2, characterized in that: The feature points in the candidate feature point sets of the second target point cloud and the source point cloud are screened respectively, including: Construct candidate feature point matrices for the second target point cloud and the source point cloud respectively, where each row in the candidate feature point matrix represents a feature point, and the pixels of the feature points are used as elements of the matrix; Calculate the average value of each column in each candidate feature point matrix; Calculate the difference between the average value of each column of each candidate feature point matrix and the preset standard value; The first N columns with the smallest absolute value of the difference are taken as the feature points in the initial feature point set, where N is a positive integer greater than 0, and N is a positive integer greater than 0.

4. The registration method of a BIM model and an engineering point cloud according to claim 2, characterized in that: After determining whether the average distance between the third target point cloud and the source point cloud is less than a preset threshold, the method further includes: If not, the process returns to the step of respectively acquiring candidate feature point sets of the second target point cloud and the source point cloud until the average distance between the third target point cloud and the source point cloud is less than a preset threshold.

5. The registration method of a BIM model and an engineering point cloud according to any one of claims 1 to 4, characterized in that: Convert the initial BIM model corresponding to the engineering scene into point cloud data, including: Store the initial BIM model in OBJ format; The initial BIM model in OBJ format is converted into PCD format to obtain the source point cloud.

6. A BIM model and engineering point cloud registration device, characterized in that: include: The engineering point cloud acquisition unit is used to acquire the point cloud data of the engineering scene and obtain the first target point cloud; A source point cloud acquisition unit is used to convert the initial BIM model corresponding to the engineering scene into point cloud data to obtain a source point cloud; A coarse registration unit, used for roughly registering the first target point cloud with the source point cloud based on pre-selected edge endpoints to obtain a second target point cloud; A precise registration unit, used for accurately registering the second target point cloud with the source point cloud to obtain a precise registration point cloud; A point cloud overlay unit is used to overlay the initial BIM model onto the precisely registered point cloud to obtain a target BIM model; Model alignment unit, used to spatially align the target BIM model with the engineering scene; The process of roughly registering the first target point cloud and the source point cloud based on the pre-selected edge endpoints by the coarse registration unit to obtain the second target point cloud includes: Determine the closest feature point and the farthest feature point in the two point clouds based on the pre-selected edge endpoints in the first target point cloud and the source point cloud respectively; Based on the edge endpoints, the nearest feature points and the farthest feature points in the point cloud, feature vector sets of the first target point cloud and the source point cloud are respectively constructed to obtain a target feature vector set and a source feature vector set; The target feature vector set and the source feature vector set are respectively transformed so that the transformed target feature vector set coincides with the source vector set, thereby obtaining a first orthogonal matrix and a second orthogonal matrix; Determine an initial point cloud transformation matrix based on the first orthogonal matrix and the second orthogonal matrix; The first target point cloud is transformed according to the initial point cloud transformation matrix to obtain a second target point cloud.

7. A BIM model and engineering point cloud registration device, characterized in that: include: Memory and processor; The memory is used to store programs; The processor is used to execute the program to implement each step of the BIM model and engineering point cloud registration method as described in any one of claims 1 to 5.

Images