Method and system for quality detection of prefabricated components based on three-dimensional computer vision

Abstract

The application discloses a prefabricated component quality detection method and system based on three-dimensional computer vision, and the method comprises the following steps: inputting point cloud data of a prefabricated component; pre-processing and contour recognizing the point cloud data; detecting the surface damage condition of the pre-processed point cloud data according to a three-dimensional target detection method to determine whether the prefabricated component meets the quality detection standard; extracting the category name, feature points and feature parameters of the recognized point cloud data; comparing the feature parameters with the design geometric dimensions of the prefabricated component to determine whether the geometric dimensions of the prefabricated component meet the quality detection standard; establishing an entity model according to the extracted category name and the corresponding feature points and feature parameters; automatically matching and establishing a finite element analysis model on the entity model; applying load and constraint conditions on the finite element analysis model to analyze the response results, and determining whether the structural performance of the prefabricated component meets the quality detection standard according to the response results. The application can improve the detection efficiency, precision and accuracy.

Description

Technical Field

[0001] This invention relates to the field of building engineering technology, and in particular to a method and system for quality inspection of prefabricated components based on three-dimensional computer vision. Background Technology

[0002] In prefabricated construction, the quality of prefabricated components (i.e., the prefabricated parts of the prefabricated building) needs to be inspected. This quality inspection mainly includes checking the apparent damage, geometric dimensions, and structural performance (including load-bearing capacity and deflection) of the prefabricated components. Currently, the quality inspection of the geometric dimensions and apparent damage of prefabricated components mainly relies on manual one-dimensional approximate linear measurements, which is time-consuming, labor-intensive, inefficient, and costly. The quality inspection of the structural performance of prefabricated components mainly relies on short-term static loading tests, non-destructive testing, and traditional finite element analysis. However, loading tests are complex and costly, non-destructive testing has low accuracy, and traditional finite element analysis methods are based on solid modeling of the prefabricated components' design dimensions, failing to consider the varying degrees of deformation that occur during processing, transportation, and installation, thus failing to guarantee the accuracy and reliability of the finite element analysis results. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for quality inspection of prefabricated components based on three-dimensional computer vision, so as to solve the problems of low efficiency and high cost of manual inspection of geometric dimensions and apparent damage, and high cost, low efficiency, low precision and poor accuracy of structural performance quality inspection.

[0004] To address the aforementioned technical problems, this invention provides a method for quality inspection of prefabricated components based on three-dimensional computer vision, comprising:

[0005] Input the point cloud data of prefabricated components for prefabricated buildings;

[0006] Preprocess the point cloud data;

[0007] The surface damage of prefabricated components is automatically detected using a 3D target detection method to determine whether the surface damage meets the quality inspection standards.

[0008] The contours of the preprocessed point cloud data are identified.

[0009] Extract the category name, feature points, and feature parameters from the point cloud data after contour recognition;

[0010] The feature parameters of the extracted point cloud data are compared with the design geometric dimensions of the precast components to determine whether the geometric dimensions of the precast components meet the quality inspection standards. The feature parameters include the geometric dimensions.

[0011] A solid model of the prefabricated component is established based on the category name of the extracted point cloud data and its corresponding feature points and feature parameters;

[0012] A finite element analysis model of the prefabricated component is automatically established on the solid model based on the category name of the extracted point cloud data and its corresponding feature points and feature parameters;

[0013] Apply loads and constraints to the finite element analysis model and analyze the response results of the finite element analysis model. Based on the response results of the finite element analysis model, determine whether the structural performance of the prefabricated component meets the quality inspection standards.

[0014] When the surface damage, geometric dimensions, and structural performance of a precast component all meet the corresponding quality inspection standards, the precast component is deemed to be of qualified quality; otherwise, it is deemed unqualified.

[0015] Furthermore, the prefabricated component quality inspection method based on three-dimensional computer vision provided by the present invention obtains point cloud data of prefabricated components of prefabricated buildings through three-dimensional laser scanning.

[0016] Furthermore, the present invention provides a method for quality inspection of prefabricated components based on three-dimensional computer vision, wherein the prefabricated components are concrete components or steel components.

[0017] Furthermore, the prefabricated component quality inspection method based on three-dimensional computer vision provided by the present invention includes the following steps for preprocessing point cloud data: filtering, denoising, and repairing the point cloud data.

[0018] Furthermore, the prefabricated component quality inspection method based on three-dimensional computer vision provided by the present invention includes the following steps to determine whether the surface damage of the prefabricated component meets the quality inspection standards:

[0019] The mean curvature of key geometric features in the preprocessed point cloud data is estimated using a quadratic surface method.

[0020] The optimal curvature threshold was obtained using the OTSU algorithm, and the damage feature region was extracted based on the threshold.

[0021] The depth information of the point cloud data is extracted, a depth mapping color model is established, and the depth information is converted. Based on the habitual location of surface damage, the color type, and the color changes between different regions, it is determined whether there is damage on the surface. If damage exists, a color threshold is set to extract the damage.

[0022] The optimal color threshold is selected from the color index values ​​corresponding to all point cloud data based on the one-dimensional maximum entropy method to reduce the damage feature area.

[0023] Based on the detected surface damage of the precast components, damage parameters are extracted to determine whether the surface damage of the precast components meets the quality inspection standards.

[0024] Furthermore, the prefabricated component quality inspection method based on three-dimensional computer vision provided by the present invention identifies the category name, feature points, and feature parameters of the preprocessed point cloud data contour.

[0025] Furthermore, the prefabricated component quality inspection method based on three-dimensional computer vision provided by this invention identifies the category names and their corresponding feature points and feature parameters as shown in the table below:

[0026]

[0027] To address the aforementioned technical problems, this invention also provides a prefabricated component quality inspection system based on three-dimensional computer vision, comprising:

[0028] The point cloud data processing unit includes:

[0029] The point cloud data input module is used to input the point cloud data of prefabricated components in prefabricated buildings;

[0030] The point cloud data preprocessing module preprocesses and performs contour recognition on the point cloud data input by the point cloud data input module.

[0031] Point cloud materialization unit, including:

[0032] The feature extraction module extracts the category name, feature points, and feature parameters of the point cloud data after contour recognition by the point cloud data preprocessing module;

[0033] The solid modeling module extracts the category names of point cloud data and their corresponding feature points and feature parameters from the feature extraction module to establish a solid model of the prefabricated component.

[0034] The point cloud quality inspection unit includes:

[0035] The surface damage detection module automatically detects the surface damage of the point cloud data after preprocessing by the point cloud data preprocessing module according to the three-dimensional target detection method to determine whether the surface damage of the precast component meets the quality inspection standard.

[0036] The geometric dimension detection module compares the feature parameters of the point cloud data extracted by the feature extraction module with the design geometric dimensions of the precast component to determine whether the geometric dimensions of the precast component meet the quality inspection standards, wherein the feature parameters include the geometric dimensions.

[0037] The structural performance inspection module automatically matches and establishes a finite element analysis model of the prefabricated component based on the category name and corresponding feature points and feature parameters extracted from the point cloud data by the feature extraction module on the solid model. Loads and constraints are applied to the finite element analysis model to analyze the response results of the finite element analysis model. Based on the response results of the finite element analysis model, it is determined whether the component performance of the prefabricated component meets the quality inspection standards.

[0038] When the surface damage detection module, geometric dimension inspection module, and structural performance inspection module in the point cloud quality inspection unit all meet the corresponding quality inspection standards, the prefabricated component is deemed to be of qualified quality; otherwise, it is deemed unqualified.

[0039] Compared with existing technologies, the beneficial effects of the prefabricated component quality inspection method and system based on three-dimensional computer vision of the present invention are as follows:

[0040] This system utilizes 3D computer vision technology to preprocess point cloud data of precast components for quality inspection of surface damage. It also identifies and extracts category names, feature points, and feature parameters from the preprocessed point cloud data to assess the geometric dimensions of the precast components. Furthermore, it uses 3D computer vision to create a solid model of the precast components by combining the extracted point cloud data with their corresponding feature points and parameters. A finite element analysis model is then automatically created on the solid model based on the extracted feature points and parameters. Finally, loads and constraints are applied to the finite element analysis model to assess the structural performance of the precast components. This fully automated and intelligent inspection of precast component quality improves efficiency, accuracy, and precision, avoids human error, and overcomes the limitations of traditional acceptance testing methods and on-site non-destructive testing.

[0041] The finite element analysis model is based on the solid model and the extracted feature points and feature parameters for automated modeling and analysis, which improves modeling efficiency and accuracy. It eliminates the tedious work of manually processing point cloud data and modeling in the traditional multi-software collaboration mode, and avoids problems such as data format incompatibility in the multi-software collaboration mode. Attached Figure Description

[0042] Figure 1 This is a flowchart of a prefabricated component quality inspection method based on 3D computer vision;

[0043] Figure 2 This is a structural diagram illustrating the compositional relationships of a prefabricated component quality inspection system based on 3D computer vision. Detailed Implementation

[0044] The present invention will now be described in detail with reference to the accompanying drawings. The advantages and features of the present invention will become clearer from the following description. It should be noted that the drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0045] Please refer to Figure 1 This invention provides a method for quality inspection of prefabricated components based on three-dimensional computer vision, comprising:

[0046] Step 101: Input the point cloud data of the prefabricated components of the prefabricated building. The prefabricated components can be concrete or steel. The point cloud data of the prefabricated components of the prefabricated building can be obtained through 3D laser scanning.

[0047] Step 102: Preprocess the point cloud data. The preprocessing steps include filtering, denoising, and repairing the point cloud data.

[0048] Step 103: The surface damage of the prefabricated component is automatically detected by the three-dimensional target detection method to determine whether the surface damage meets the quality inspection standards.

[0049] Step 104: Identify the contours of the preprocessed point cloud data. The identification includes category names, feature points, and feature parameters.

[0050] Step 105: Extract the category name, feature points, and feature parameters of the point cloud data after contour recognition.

[0051] Step 106: Compare the feature parameters of the extracted point cloud data with the design geometric dimensions of the precast components to determine whether the geometric dimensions of the precast components meet the quality inspection standards. The feature parameters include the geometric dimensions.

[0052] Step 107: Establish a solid model of the prefabricated component based on the category name of the extracted point cloud data and its corresponding feature points and feature parameters.

[0053] Step 108: Automatically match and establish a finite element analysis model of the prefabricated component on the solid model based on the category name of the extracted point cloud data and its corresponding feature points and feature parameters.

[0054] Step 109: Apply loads and constraints to the finite element analysis model and analyze the response results of the finite element analysis model. Based on the response results of the finite element analysis model, determine whether the structural performance of the prefabricated component meets the quality inspection standards.

[0055] Step 110: When the surface damage, geometric dimensions and structural performance of the precast component meet the corresponding quality inspection standards, the quality of the precast component is deemed qualified; otherwise, it is deemed unqualified.

[0056] The precast component quality inspection method based on three-dimensional computer vision provided in this embodiment of the invention can identify the category names of standard precast components and their corresponding feature points and feature parameters as shown in Table 1 below:

[0057]

[0058] Table 1

[0059] Please refer to Figure 2 This invention also provides a precast component quality inspection system based on three-dimensional computer vision. Based on the aforementioned precast component quality inspection method, it includes a point cloud data processing unit, a point cloud solidification unit, a point cloud quality inspection unit, and a structural performance inspection unit, wherein:

[0060] The point cloud data processing unit includes:

[0061] The point cloud data input module is used to input the point cloud data of prefabricated components in prefabricated buildings.

[0062] The point cloud data preprocessing module performs preprocessing and contour recognition on the point cloud data input by the point cloud data input module.

[0063] Point cloud materialization unit, including:

[0064] The feature extraction module extracts the category name, feature points, and feature parameters of the point cloud data after contour recognition by the point cloud data preprocessing module.

[0065] The solid modeling module extracts the category names and corresponding feature points and feature parameters of the point cloud data from the feature extraction module to establish a solid model of the prefabricated component.

[0066] The point cloud quality inspection unit includes:

[0067] The surface damage detection module automatically detects the surface damage of the point cloud data after preprocessing by the point cloud data preprocessing module according to the three-dimensional target detection method to determine whether the surface damage of the precast component meets the quality inspection standards.

[0068] The geometric dimension detection module compares the feature parameters of the point cloud data extracted by the feature extraction module with the design geometric dimensions of the precast component to determine whether the geometric dimensions of the precast component meet the quality inspection standards, wherein the feature parameters include the geometric dimensions.

[0069] The structural performance inspection module automatically matches and establishes a finite element analysis model of the precast component based on the category names and corresponding feature points and feature parameters extracted from the point cloud data by the feature extraction module on the solid model. Loads and constraints are applied to the finite element analysis model, and the response results are analyzed. Based on the response results of the finite element analysis model, it is determined whether the performance of the precast component meets the quality inspection standards; that is, according to current relevant specifications, based on the response results of the finite element analysis of the precast component, it is determined whether the structural performance of the precast component meets the acceptance standards for precast components.

[0070] When the surface damage detection module, geometric dimension inspection module, and structural performance inspection module in the point cloud quality inspection unit all meet the corresponding quality inspection standards, the prefabricated component is deemed to be of qualified quality; otherwise, it is deemed unqualified.

[0071] The structural performance verification module performs load-bearing capacity analysis and calculations on the input standard precast component solid model, calculating the response results. Specifically, firstly, based on the material properties input from the surface damage assessment module, it determines the precast component material model, element type (including rod elements, plate elements, solid elements, etc.), load form (including displacement constraints, concentrated forces, surface loads, volume loads, inertial forces, coupled field loads, etc.), analysis type (including static analysis, modal analysis, harmonic response analysis, buckling analysis, dynamic analysis, etc.), and post-processing parameters. Next, it adaptively meshes the reconstructed solid model to form a finite element mesh. Then, it selects the boundary conditions and coupling relationships between degrees of freedom of the standard precast component, applying loads to the finite element model (key points, lines, surfaces) or (nodes, elements). Further, based on the actual solution requirements, it specifies the appropriate analysis type and performs finite element calculations. Finally, it extracts the response analysis data (including stress, strain, displacement, etc.) of the standard precast component.

[0072] The present invention provides a method and system for quality inspection of precast components based on three-dimensional computer vision. This method utilizes three-dimensional computer vision technology to preprocess point cloud data of the input precast components to detect surface damage. It then uses this technology to identify and extract category names, feature points, and feature parameters from the preprocessed point cloud data to inspect the geometric dimensions of the precast components. Finally, it uses three-dimensional computer vision technology to create a solid model of the precast components by combining the extracted point cloud data with their corresponding feature points and parameters. Based on the extracted feature points and parameters, a finite element analysis model is automatically created on the solid model. By applying loads and constraints to the finite element analysis model, the structural performance of the precast components is inspected. This achieves fully automated and intelligent quality inspection of precast components, improving inspection efficiency, accuracy, and precision, avoiding human error, and overcoming the limitations of the randomness of acceptance tests and the limitations of on-site non-destructive testing in traditional inspection methods.

[0073] The prefabricated component quality inspection method and system based on three-dimensional computer vision provided by this invention features an automated modeling and analysis of the finite element analysis model based on the solid model and extracted feature points and feature parameters. This improves modeling efficiency and accuracy, eliminates the tedious manual processing of point cloud data and modeling work in the traditional multi-software collaboration mode, and avoids problems such as data format incompatibility in the multi-software collaboration mode.

[0074] The prefabricated component quality inspection method and system based on three-dimensional computer vision provided in this invention can simultaneously inspect one or more prefabricated components of one or more categories, thereby improving inspection efficiency.

[0075] To assess surface damage in precast components, the present invention provides a precast component quality inspection method and system based on three-dimensional computer vision. The method for determining whether surface damage in precast components meets quality inspection standards includes the following steps:

[0076] Step 301: The mean curvature of the key (important) geometric features of the preprocessed point cloud data is estimated using the quadratic surface method. Specifically, first, the equation of the spatial quadratic surface passing through a point p0 in the point cloud data and its k-neighborhood point set is obtained; then, the equations of the normal vector passing through point p0 and the two mutually perpendicular radial planes perpendicular to the tangent plane of point p0 are obtained; next, the intersection spatial curve and the curvature of point p0 on the curve are obtained using the quadratic surface equation and the equations of the two radial planes intersecting point p0; finally, the mean of the curvatures of the two orthogonal curves at point p0 is the estimated mean curvature.

[0077] Step 302: The optimal curvature threshold is obtained using the OTSU algorithm (Otsu's method—maximum inter-class variance method), and the damage feature region is extracted based on the threshold. Specifically, first, an initial curvature threshold δ is set, and the point cloud is divided into two classes, C1 and C2, and the mean curvature values ​​μ1 and μ2 of the two classes and the mean curvature value μ of the overall point cloud are calculated; then, based on the probabilities P1 and P2 of the two classes of point clouds... Calculate the between-class variance Then, the optimal threshold is obtained. And set it as the average curvature threshold for feature point cloud data extraction. If the average curvature value H of the point cloud... i If the value is greater than or equal to σ, it is determined to be a feature point of the damaged area. This feature point is then extracted to achieve damage detection and damage area localization.

[0078] Step 303: Extract depth information from point cloud data, establish a depth-mapping color model, and realize the conversion of depth information. Specifically, first, a reference plane is set. To reduce the deviation of depth information caused by the tilt of the experimental platform or the workpiece itself, the least squares method is used to fit the 3D point cloud data to obtain the plane ax+by+cz+d=0 corresponding to the surface of the prefabricated component. This plane is used as the measurement reference plane to calculate the depth hi of each point cloud, simplifying data processing and normalizing the depth information. The point cloud depth information is then mapped into color component information. The 3D point cloud data is converted into a corresponding 2D color point set on the XY plane, including color information. Based on the habitual location of surface damage, color type, and color changes between different areas, it is determined whether there is damage on the surface. If damage exists, a color threshold is set, and damage extraction is performed.

[0079] Step 304: Based on the one-dimensional maximum entropy method, select the optimal color threshold from the color index values ​​corresponding to all point cloud points to narrow down the damage feature region of step 302. Specifically, first, let ε be the color threshold, and divide the two-dimensional color point set into regions A and B. Let the points with color index values ​​of 1 to ε and the points with color index values ​​of ε+1 to L belong to regions A and B respectively, then the probability distributions of regions A and B are as follows: and P B =1-P(ε), where L is the maximum color index value; then, define the entropy related to the probability distributions of regions A and B. (in, L = 64); Finally, the ε that maximizes the criterion function φ(ε) = H(A) + H(B) is the optimal threshold. For different damage types, regions A and B are set differently. If regions A and B are normal and damaged regions respectively, then N is extracted. RGB For point cloud data greater than ε, if regions A and B are damaged and normal regions respectively, then extract N. RGB Point cloud data < ε.

[0080] Step 305: Based on the detected surface damage of the precast component, extract damage parameters, including the average value of the damage width and depth difference, and set the reduced material properties of the damaged area according to the damage parameters; so as to determine whether the surface damage of the precast component meets the quality inspection standards based on the damage parameters.

[0081] This invention is not limited to the specific embodiments described above. Obviously, the embodiments described above are only a part of the embodiments of this invention, not all of them. All other embodiments obtained by those skilled in the art based on the described embodiments of this invention are within the scope of protection of this invention. Those skilled in the art can make other modifications and variations to this invention. Therefore, if these modifications and variations of this invention fall within the scope of the claims of this invention, then this invention also intends to include these modifications and variations.

Claims

1. A method for quality inspection of prefabricated components based on three-dimensional computer vision, characterized in that, include: Input the point cloud data of prefabricated components for prefabricated buildings; Preprocess the point cloud data; The surface damage of prefabricated components is automatically detected using a 3D target detection method to determine whether the surface damage meets the quality inspection standards. The contours of the preprocessed point cloud data are identified. Extract the category name, feature points, and feature parameters from the point cloud data after contour recognition; The feature parameters of the extracted point cloud data are compared with the design geometric dimensions of the precast components to determine whether the geometric dimensions of the precast components meet the quality inspection standards. The feature parameters include the geometric dimensions. A solid model of the prefabricated component is established based on the category name of the extracted point cloud data and its corresponding feature points and feature parameters; A finite element analysis model of the prefabricated component is automatically established on the solid model based on the category name of the extracted point cloud data and its corresponding feature points and feature parameters; Apply loads and constraints to the finite element analysis model and analyze the response results of the finite element analysis model. Based on the response results of the finite element analysis model, determine whether the structural performance of the prefabricated component meets the quality inspection standards. When the surface damage, geometric dimensions, and structural performance of a precast component all meet the corresponding quality inspection standards, the precast component is deemed to be of qualified quality; otherwise, it is deemed unqualified.

2. The method for quality inspection of prefabricated components based on three-dimensional computer vision according to claim 1, characterized in that, Point cloud data of prefabricated components for prefabricated buildings are obtained through 3D laser scanning.

3. The method for quality inspection of prefabricated components based on three-dimensional computer vision according to claim 1, characterized in that, The precast components are either concrete or steel components.

4. The method for quality inspection of prefabricated components based on three-dimensional computer vision according to claim 1, characterized in that, The preprocessing steps for point cloud data include filtering, denoising, and repairing.

5. The method for quality inspection of prefabricated components based on three-dimensional computer vision according to claim 1, characterized in that, The method for determining whether the surface damage of precast components meets the quality inspection standards includes the following steps: The mean curvature of key geometric features in the preprocessed point cloud data is estimated using a quadratic surface method. The optimal curvature threshold was obtained using the OTSU algorithm, and the damage feature region was extracted based on the threshold. Extract depth information from point cloud data, establish a depth mapping color model to realize the conversion of depth information, and determine whether there is damage on the surface based on the habitual occurrence location of surface damage, color type, and color changes between different regions. If damage exists, set a color threshold and extract the damage. The optimal color threshold is selected from the color index values ​​corresponding to all point cloud data based on the one-dimensional maximum entropy method to reduce the damage feature area. Based on the detected surface damage of the precast components, damage parameters are extracted to determine whether the surface damage of the precast components meets the quality inspection standards.

6. The method for quality inspection of prefabricated components based on three-dimensional computer vision according to claim 1, characterized in that, The content for contour recognition of preprocessed point cloud data includes category name, feature points, and feature parameters.

7. The method for quality inspection of prefabricated components based on three-dimensional computer vision according to claim 6, characterized in that, The identified category names and their corresponding feature points and feature parameters are shown in the table below: 。 8. A prefabricated component quality inspection system based on three-dimensional computer vision, characterized in that, include: The point cloud data processing unit includes: The point cloud data input module is used to input the point cloud data of prefabricated components in prefabricated buildings; The point cloud data preprocessing module preprocesses and performs contour recognition on the point cloud data input by the point cloud data input module. Point cloud materialization unit, including: The feature extraction module extracts the category name, feature points, and feature parameters of the point cloud data after contour recognition by the point cloud data preprocessing module; The solid modeling module extracts the category names of point cloud data and their corresponding feature points and feature parameters from the feature extraction module to establish a solid model of the prefabricated component. The point cloud quality inspection unit includes: The surface damage detection module automatically detects the surface damage of the point cloud data after preprocessing by the point cloud data preprocessing module according to the three-dimensional target detection method to determine whether the surface damage of the precast component meets the quality inspection standard. The geometric dimension detection module compares the feature parameters of the point cloud data extracted by the feature extraction module with the design geometric dimensions of the precast component to determine whether the geometric dimensions of the precast component meet the quality inspection standards, wherein the feature parameters include the geometric dimensions. The structural performance inspection module automatically matches and establishes a finite element analysis model of the prefabricated component based on the category name and corresponding feature points and feature parameters extracted from the point cloud data by the feature extraction module on the solid model. Loads and constraints are applied to the finite element analysis model to analyze the response results of the finite element analysis model. Based on the response results of the finite element analysis model, it is determined whether the component performance of the prefabricated component meets the quality inspection standards. When the surface damage detection module, geometric dimension inspection module, and structural performance inspection module in the point cloud quality inspection unit all meet the corresponding quality inspection standards, the precast component is deemed to be of qualified quality; otherwise, it is deemed unqualified.

Images