Line heating forming deviation measuring method, system and equipment based on three-dimensional point cloud

Abstract

The invention provides a line heating forming deviation measuring method, system and equipment based on three-dimensional point cloud. The line heating forming deviation measuring method comprises the steps that ship plate measuring point cloud data is acquired through scanning equipment, and ship plate rib line sparse feature point data and a ship plate CAD design file are obtained through ship plate design software; performing curve fitting interpolation on the sparse feature points of the rib lines of the ship plate by using a B-spline curve fitting algorithm; the CAD file is converted into an OBJ format from an STP format, and then the OBJ file becomes theoretical model point cloud through a triangular patch point cloud interpolation algorithm; carrying out voxel size self-adaptive down-sampling preprocessing on the ship plate measurement point cloud data obtained by scanning; carrying out improved coarse registration and fine registration; fitting a plane where the theoretical rib line point cloud is located, utilizing the plane to extract and measure a ship plate rib line, and calculating a rib line forming deviation; searching a point closest to the theoretical point cloud in the measurement point cloud, and displaying the integral forming deviation of the ship plate through a color difference cloud picture. According to the invention, the accurate measurement of the line heating forming deviation is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of 3D point cloud measurement, and in particular to a method, system and device for measuring water-fire bending plate forming deviation based on three-dimensional point cloud. Background Art

[0002] The water-fire bending process has long been widely used in the processing and manufacturing of complex curved surfaces of hull outer plates, and the automatic detection and identification of these curved surfaces is the key technology to realize the automated processing of hull bending. At present, shipyard production mainly relies on the experience of skilled workers and uses samples or sample boxes for detection. This manual experience-based detection method has low efficiency and large errors, and cannot meet the requirements of modern shipbuilding industry for automated production. It has become a bottleneck restricting the production cycle and quality of ships. Therefore, realizing the automated identification of the curved surfaces of hull outer plates is a difficult problem that needs to be solved urgently, and it is also a key component of the automated processing of water-fire bending.

[0003] The traditional two-dimensional detection method is time-consuming and laborious to detect shape and size, and some complex curved surfaces are difficult to detect with high precision. Its detection accuracy and efficiency have been difficult to meet the detection standards and requirements of modern industrial manufacturing technology for various parts. Modern digital detection technology based on three-dimensional CAD models, that is, using three-dimensional CAD models to compare with three-dimensional point cloud measurement models, is gradually replacing traditional detection technology and is being used more and more widely. It has become the core technology for the detection of various high-precision complex parts. Among them, the registration technology of three-dimensional measurement point clouds and CAD models is the basis and core technology of three-dimensional point cloud digital detection technology, and it is also the core technology in water and fire bending ship plate detection technology. Based on this, there is an urgent need for a water and fire bending plate forming deviation measurement method, system and equipment based on three-dimensional point clouds. Summary of the invention

[0004] In order to solve the above problems, the present invention aims to propose a method, system and equipment for measuring the forming deviation of water-fire bending plate based on three-dimensional point cloud, so as to realize the accurate measurement of the forming deviation of water-fire bending plate.

[0005] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0006] A method for measuring the forming deviation of a water-fire bending plate based on a three-dimensional point cloud comprises the following steps:

[0007] S1: Use scanning equipment to collect the measured point cloud data of the ship plate, and use the ship plate design software to obtain the sparse feature point data of the ship plate rib line and the ship plate CAD design file;

[0008] S2: Use the B-spline curve fitting algorithm to perform curve fitting on the sparse feature point data of the ship plate rib line and interpolate the rib line point cloud based on the feature points;

[0009] S3: Use the OCC geometry library to convert the shipboard CAD design file from STP format to OBJ format, and then use the triangular patch point cloud interpolation algorithm on the surface of the OBJ model file to make it a theoretical model point cloud;

[0010] S4: Preprocessing the measured point cloud data of the shipboard obtained by actual scanning using a voxel size adaptive downsampling algorithm;

[0011] S5: The improved coarse registration algorithm is used to coarsely register the shipboard measurement point cloud and the theoretical model point cloud, and then the ICP registration algorithm is used for fine registration;

[0012] S6: The plane where the theoretical ship plate rib line is located is obtained by using the RANSAC plane fitting algorithm, and the measured ship plate point cloud surface is intercepted by the plane to obtain the measured ship plate rib line point cloud, and the error between the two is calculated, which is the ship plate rib line forming deviation; the nearest point to the theoretical point cloud is searched in the measured point cloud, and the error between the two points of the ship plate measured point cloud and the theoretical model point cloud is calculated, and the overall forming deviation of the ship plate is displayed through the color difference cloud map.

[0013] Furthermore, the step S2 specifically includes:

[0014] Step S21: storing each rib line in a container according to the rib line number segment in the PCD point cloud file of the ship plate rib line feature points;

[0015] Step S22: performing B-spline curve fitting on each rib feature point in the container;

[0016] Step S23: performing point cloud sampling interpolation on the fitted rib curve;

[0017] Furthermore, the step S3 specifically includes:

[0018] Step S31: read the STP file through the OCC geometry core library and convert the STP file into OBJ format;

[0019] Step S32: Use the Point Cloud Open Source Library (PCL) to read the OBJ file and convert the data into a data format that can be processed by PCL;

[0020] Step S33: traverse all the triangles in the mesh file, obtain the three vertices O, A, and B of the triangle, calculate the vectors of the two edges OA and OB, and calculate the area S of the triangle through the vectors OA and OB. The formula is:

[0021]

[0022] Step S34: derive the number of points to be inserted, generate a total number of points N, and insert density Den, then:

[0023] N=S×Den

[0024] Step S35: processing the decimal part of the generated points, and rounding the generated points to an integer using the principle of rounding off;

[0025] Step S36: After obtaining the number of generated points, interpolation is performed inside the triangle patch. First, two random numbers x∈(0,1) and y∈(0,1) are generated. Then, it is determined whether x+y is greater than 1. If it is greater than 1, it means that the generated point will be outside the AB edge, that is, the generated point is not inside the triangle. Then, x and y are processed to make x=1-x, y=1-y, that is, the point cloud is generated inside the triangle.

[0026] Step S37: Generate points O is a vertex of a triangle, x and y are parameters used to control the offset of P in the OA and OB directions. The generated point P is placed in the point cloud container until all triangular faces are traversed.

[0027] Furthermore, the step S4 specifically includes:

[0028] Step S41: traverse the entire point cloud and calculate the average Euclidean distance Δd between neighboring points of the entire point cloud;

[0029] Step S42: The average Euclidean distance calculated in step S41 is enlarged by n times, ie, nΔd, as the input of the voxel size in the voxel filtering downsampling algorithm, so as to realize the voxel size adaptive downsampling of the measured ship plate point cloud data.

[0030] Furthermore, the step S5 specifically includes:

[0031] Step S51: Calculate the covariance matrix and centroid C of the pre-processed shipboard measurement point cloud data S , solve its eigenvector U through the covariance matrix S (i), where i = 1, 2, 3, which represent the main directions of the three axes x, y, and z, respectively. The new matrix U is constructed by the three eigenvectors;

[0032] Step S52: Calculate the covariance matrix and centroid C of the theoretical point cloud data of the shipboard t , solve its eigenvector U through the covariance matrix t (i), where i = 1, 2, 3, a new matrix V is constructed from three eigenvectors;

[0033] Step S53: Solve the rotation matrix R and translation matrix T, where:

[0034] R=UV T

[0035] T=C S-RC t

[0036] Step S54: roughly aligning the theoretical point cloud data of the shipboard with the point cloud data of the measurement board through a rotation and translation matrix;

[0037] Step S55: Calculate the RMS root mean square error E between the two point clouds after the rough registration is completed r , that is, the registration error calculation;

[0038] Step S56: Transform the feature vector U S (i) Take the inverse, that is, 2 3 In each of the eight cases, the root mean square error E is discussed. r , repeat steps S4.1-S4.5 to find E r The corresponding rotation matrix R and translation matrix T when the minimum is obtained are the results of rough registration;

[0039] Step S57: Finally, the measured plate point cloud and the theoretical ship plate point cloud are precisely registered by ICP based on SVD decomposition;

[0040] Step S58: First, use the exhaustive search method to measure each point p in the shipboard point cloud P. i , find and p in the theoretical ship plate point cloud Q i The point q that is closest to the point i ,q i As p i The corresponding points of form a total of n corresponding point pairs;

[0041] Step S59: Minimize the objective function value of the formula f(R, T), and solve the rotation matrix R and the translation matrix T through SVD decomposition;

[0042]

[0043] Step S510: using the obtained transformation matrix to update the source point cloud P, obtaining a new point cloud P′, and using the new point cloud as a new measurement ship plate point cloud;

[0044] P′=RP+T

[0045] Step S511: Repeat the process of step S57 to step S510 until the iteration condition is met, and the iteration condition is to reach the maximum number of iterations K max , or the root mean square error d between iterative corresponding points k Or the root mean square error of two iterations d k -d k-1 Less than a certain value;

[0046]

[0047] Step S512: The rotation and translation parameters obtained when the iteration is stopped are used as the optimal transformation matrix. According to the parameters, the measured ship plate point cloud is transformed into the theoretical ship plate point cloud, and finally the point cloud registration is completed.

[0048] Furthermore, the step S6 specifically includes:

[0049] Step S61: select a theoretical ship plate rib line, randomly select 3 points, and determine whether they are collinear. If they are collinear, reselect them. If they are not collinear, calculate the plane parameter equation Ax+By+Cz+D=0 through the selected 3 points, where A, B, C are the plane normal vector components, and D is the plane offset;

[0050] Step S62: For each point in the theoretical ship plate rib line, calculate the point (x 0 ,y 0 ,z 0 ) to the fitting plane, the calculation formula is as follows:

[0051]

[0052] If the distance is less than the set threshold, it is considered an "inside point", and if it exceeds the threshold, it is considered an "outside point";

[0053] Step S63: Count the number of all inliers, which are points that fit the current plane model. Repeat steps 61-62 multiple times, and each time randomly select 3 points from the data set to fit the plane, calculate the inliers, and record the number of inliers. If the number of inliers obtained in a certain iteration exceeds the previous maximum number of inliers, save the current plane model.

[0054] Step S64: repeating steps 61 to 63 until the plane fitting of each theoretical ship plate rib line is completed;

[0055] Step S65: intercepting the rib lines in the measured ship plate point cloud through these fitting planes, that is, traversing the measured ship plate point cloud, substituting its coordinates into the fitting plane equation, and if its coordinates conform to the plane equation, it is regarded as the measured ship plate rib line, so as to extract the measured ship plate rib line;

[0056] Step S66: After the plate rib line extraction is completed, the error between the two is calculated, which is the plate rib line forming deviation;

[0057] Step S67: Search the measured point cloud for the point closest to the theoretical point cloud, compare the theoretical model point cloud and the measured ship plate point cloud, calculate the Euclidean distance between the two, form a color difference cloud map, and display the overall forming deviation of the ship plate.

[0058] Furthermore, the scanning device in step S1 is a 3D camera.

[0059] In order to achieve the above object, the present invention also provides a water-fire bending plate forming deviation measurement system based on three-dimensional point cloud, comprising the following modules:

[0060] Scanning module: Use scanning equipment to collect the measured point cloud data of the ship plate, and use the ship plate design software to obtain the sparse feature point data of the ship plate rib line and the ship plate CAD design file;

[0061] Fitting module: Use B-spline curve fitting algorithm to perform curve fitting on the sparse feature point data of the ship plate rib line and interpolate the rib line point cloud based on the feature points;

[0062] Conversion module: Use the OCC geometry library to convert the shipboard CAD design file from STP format to OBJ format, and then use the triangular patch point cloud interpolation algorithm on the surface of the OBJ model file to make it a theoretical model point cloud;

[0063] Preprocessing module: used to preprocess the shipboard measurement point cloud data obtained by actual scanning using a voxel size adaptive downsampling algorithm;

[0064] Registration module: used to roughly register the shipboard measurement point cloud and the theoretical model point cloud using the improved rough registration algorithm, and then use the ICP registration algorithm for fine registration;

[0065] Forming deviation measurement module: The plane where the theoretical ship plate rib line is located is obtained by using the RANSAC plane fitting algorithm, and the rib line on the measured ship plate is intercepted by this plane, the error between the two is calculated, and the forming deviation of the ship plate rib line is determined; the nearest point to the theoretical point cloud is searched in the measured point cloud, the error between the ship plate measured point cloud and the theoretical model point cloud is calculated, and the overall forming deviation of the ship plate is determined by the color difference cloud map.

[0066] In order to achieve the above-mentioned objectives, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the water-fire bending plate forming deviation measurement method based on three-dimensional point cloud as described above is implemented.

[0067] Beneficial effects: The present invention uses a B-spline fitting algorithm to fit and interpolate the rib line feature point data, thereby ensuring the integrity of the ship plate rib line point cloud data; uses a triangular patch interpolation algorithm to convert the ship plate STP file into a point cloud, thereby facilitating subsequent global error calculation; uses a voxel size adaptive downsampling algorithm to downsample the measured ship plate point cloud, thereby reducing its data volume and accelerating the calculation; uses an improved coarse registration algorithm and a fine registration algorithm to align the measured ship plate with the theoretical ship plate, thereby laying the foundation for subsequent error calculation; innovatively uses the RANSAC plane fitting algorithm to fit the plane where the theoretical ship plate rib line is located, thereby intercepting the rib line point cloud on the measured ship plate, thereby calculating the rib line error and the global error. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0069] Figure 1 It is a flow chart of the method for measuring the forming deviation of a water-fire bending plate based on a three-dimensional point cloud according to an embodiment of the present invention;

[0070] Figure 2 The ship plate STP model file and theoretical rib line feature point data of the water-fire bending plate forming deviation measurement method based on three-dimensional point cloud described in the embodiment of the present invention;

[0071] Figure 3 The ship plate measurement point cloud of the water-fire bending plate forming deviation measurement method based on three-dimensional point cloud according to the embodiment of the present invention;

[0072] Figure 4 The point cloud obtained by fitting and interpolating the rib line feature points using the B-spline fitting algorithm in the method for measuring the deviation of water-fire bending plate based on three-dimensional point cloud described in an embodiment of the present invention;

[0073] Figure 5 It is an algorithm flow chart of converting the STP file into point cloud of the method for measuring the deviation of water-fire bending plate forming based on three-dimensional point cloud according to an embodiment of the present invention;

[0074] Figure 6 A flow chart of a voxel size adaptive downsampling algorithm of a method for measuring deviation of water-fire plate bending based on three-dimensional point cloud according to an embodiment of the present invention;

[0075] Figure 7 A flow chart of a registration algorithm for a method for measuring deviations of water-fire plate bending based on three-dimensional point clouds according to an embodiment of the present invention;

[0076] Figure 8 It is a flow chart of the algorithm for extracting the rib line of a ship plate according to the method for measuring the deviation of a water-fire bending plate based on three-dimensional point cloud according to an embodiment of the present invention;

[0077] Fig. 9 This is a rendering of the ship plate rib line extraction effect of the water-fire bending plate forming deviation measurement method based on three-dimensional point cloud according to an embodiment of the present invention;

[0078] Fig.10 This is a rendering of a global error cloud diagram of a ship plate of a method for measuring deviations of water-fire bending plate forming based on three-dimensional point clouds according to an embodiment of the present invention;

[0079] Fig.11It is a structural schematic diagram of the water-fire bending plate forming deviation measurement system based on three-dimensional point cloud according to an embodiment of the present invention. DETAILED DESCRIPTION

[0080] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0081] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0082] Generally speaking, the measured surface and the designed surface are in different coordinate systems. To evaluate their errors, the two coordinate systems must be rigidly transformed into one coordinate system for matching. Therefore, the theoretical data of the designed surface used in the present invention are the characteristic point data of the ship plate rib line and the ship plate CAD design file, and the rib line shape deviation and the overall forming deviation of the ship plate need to be judged. Since the number of point clouds of the characteristic points of the ship plate rib line is too small, it is necessary to perform curve fitting and interpolation to obtain the complete ship plate rib line, and then align it with the measurement plate to calculate the error; the ship plate CAD design file also has the same problem, so it is necessary to convert the CAD model file and interpolate it into three-dimensional point cloud data, and then align it with the measurement plate to calculate the error.

[0083] Example 1

[0084] Based on the above design concept, see Figure 1-10 :A method for measuring the forming deviation of a water-fire bending plate based on a three-dimensional point cloud in this embodiment includes the following steps:

[0085] S1: Use scanning equipment to collect the measured point cloud data of the ship plate, and use the ship plate design software to obtain the sparse feature point data of the ship plate rib line and the ship plate CAD design file;

[0086] S2: Use the B-spline curve fitting algorithm to perform curve fitting on the sparse feature point data of the ship plate rib line and interpolate the rib line point cloud based on the feature points;

[0087] S3: Use the OCC geometry library to convert the shipboard CAD design file from STP format to OBJ format, and then use the triangular patch point cloud interpolation algorithm on the surface of the OBJ model file to make it a theoretical model point cloud;

[0088] S4: Preprocessing the measured point cloud data of the shipboard obtained by actual scanning using a voxel size adaptive downsampling algorithm;

[0089] S5: The coarse registration algorithm improved for the direction vector is used to coarsely register the measured point cloud of the ship plate and the theoretical model point cloud, and then the ICP registration algorithm is used for fine registration;

[0090] S6: The plane where the theoretical ship plate rib line is located is obtained by using the RANSAC plane fitting algorithm, and the measured ship plate point cloud surface is intercepted by the plane to obtain the measured ship plate rib line point cloud, and the error between the two is calculated, which is the ship plate rib line forming deviation; the nearest point to the theoretical point cloud is searched in the measured point cloud, and the error between the ship plate measured point cloud and the theoretical model point cloud is calculated, and the overall forming deviation of the ship plate is displayed through the color difference cloud map.

[0091] This embodiment uses the B-spline fitting algorithm to fit and interpolate the rib line feature points, thereby ensuring the integrity of the ship plate rib line point cloud data; uses the triangular patch interpolation algorithm to convert the ship plate STP file into a point cloud, thereby facilitating the subsequent global error calculation; uses the voxel size adaptive downsampling algorithm to downsample the measured ship plate point cloud, thereby reducing its data volume and accelerating the calculation; uses the improved coarse registration algorithm and fine registration algorithm to align the measured ship plate with the theoretical ship plate, thereby laying the foundation for the subsequent error calculation; innovatively uses the RANSAC plane fitting algorithm to fit the plane where the theoretical ship plate rib line is located, thereby intercepting the rib line point cloud on the measured ship plate, thereby calculating the rib line error and the global error.

[0092] In a specific example, the step S2 specifically includes:

[0093] Step S21: storing each rib line in a container according to the rib line number segment in the PCD point cloud file of the ship plate rib line feature points;

[0094] Step S22: performing B-spline curve fitting on each rib feature point in the container;

[0095] Step S23: performing point cloud sampling interpolation on the fitted rib curve;

[0096] This embodiment uses the B-spline fitting algorithm to fit and interpolate the rib line feature points, thereby ensuring the integrity of the ship plate rib line point cloud data.

[0097] In a specific example, step S3 specifically includes:

[0098] Step S31: read the STP file through the OCC geometry core library and convert the STP file into OBJ format;

[0099] Step S32: Use the Point Cloud Open Source Library (PCL) to read the OBJ file and convert the data into a data format that can be processed by PCL;

[0100] Step S33: traverse all the triangles in the mesh file, obtain the three vertices O, A, and B of the triangle, calculate the vectors of the two edges OA and OB, and calculate the area S of the triangle through the vectors OA and OB. The formula is:

[0101]

[0102] Step S34: derive the number of points to be inserted, generate a total number of points N, and insert density Den, then:

[0103] N=S×Den

[0104] Step S35: processing the decimal part of the generated points, and rounding the generated points to an integer using the principle of rounding off;

[0105] Step S36: After obtaining the number of generated points, interpolation is performed inside the triangle patch. First, two random numbers x∈(0,1) and y∈(0,1) are generated. Then, it is determined whether x+y is greater than 1. If it is greater than 1, it means that the generated point will be outside the AB edge, that is, the generated point is not inside the triangle. Then, x and y are processed to make x=1-x, y=1-y, that is, the point cloud is generated inside the triangle.

[0106] Step S37: Generate points O is a vertex of a triangle, x and y are parameters used to control the offset of P in the OA and OB directions. The generated point P is placed in the point cloud container until all triangular faces are traversed.

[0107] This embodiment uses a triangular patch interpolation algorithm to convert the ship plate STP file into a point cloud to facilitate subsequent global error calculation.

[0108] In a specific example, the step S4 specifically includes:

[0109] Step S41: traverse the entire point cloud and calculate the average Euclidean distance Δd between neighboring points of the entire point cloud;

[0110] Step S42: The average Euclidean distance calculated in step S41 is enlarged by n times, ie, nΔd, as the input of the voxel size in the voxel filtering downsampling algorithm, so as to realize the voxel size adaptive downsampling of the measured ship plate point cloud data.

[0111] This embodiment uses a voxel size adaptive downsampling algorithm to downsample the measured ship plate point cloud, thereby reducing its data volume and accelerating calculation.

[0112] In a specific example, the step S5 specifically includes:

[0113] Step S51: Calculate the covariance matrix and centroid C of the pre-processed shipboard measurement point cloud data S , solve its eigenvector U through the covariance matrix S (i), where i = 1, 2, 3, which represent the main directions of the three axes x, y, and z, respectively. The new matrix U is constructed by the three eigenvectors;

[0114] Step S52: Calculate the covariance matrix and centroid C of the theoretical point cloud data of the shipboard t , solve its eigenvector U through the covariance matrix t (i), where i = 1, 2, 3, a new matrix V is constructed from three eigenvectors;

[0115] Step S53: Solve the rotation matrix R and translation matrix T, where:

[0116] R=UV T

[0117] T=C S -RC t

[0118] Step S54: roughly aligning the theoretical point cloud data of the shipboard with the point cloud data of the measurement board through a rotation and translation matrix;

[0119] Step S55: Calculate the RMS root mean square error E between the two point clouds after the rough registration is completed r , that is, the registration error calculation;

[0120] Step S56: Transform the feature vector U S (i) Take the inverse, that is, 2 3 In each of the eight cases, the root mean square error E is discussed. r , repeat steps S4.1-S4.5 to find E r The corresponding rotation matrix R and translation matrix T when the minimum is obtained are the results of rough registration;

[0121] Step S57: Finally, the measured plate point cloud and the theoretical ship plate point cloud are precisely registered by ICP based on SVD decomposition;

[0122] Step S58: First, use the exhaustive search method to measure each point p in the shipboard point cloud P. i , find and p in the theoretical ship plate point cloud Q i The point q that is closest to the point i ,q i As p i The corresponding points of form a total of n corresponding point pairs;

[0123] Step S59: Minimize the objective function value of the formula f(R, T), and solve the rotation matrix R and the translation matrix T through SVD decomposition;

[0124]

[0125] Step S510: using the obtained transformation matrix to update the source point cloud P, obtaining a new point cloud P′, and using the new point cloud as a new measurement ship plate point cloud;

[0126] P′=RP+T

[0127] Step S511: Repeat the process of step S57 to step S510 until the iteration condition is met, and the iteration condition is to reach the maximum number of iterations K max , or the root mean square error d between iterative corresponding points k Or the root mean square error of two iterations d k -d k-1 Less than a certain value;

[0128]

[0129] Step S512: The rotation and translation parameters obtained when the iteration is stopped are used as the optimal transformation matrix. According to the parameters, the measured ship plate point cloud is transformed into the theoretical ship plate point cloud, and finally the point cloud registration is completed.

[0130] This embodiment uses an improved coarse registration algorithm and a fine registration algorithm to align the measured ship plate with the theoretical ship plate, laying a foundation for subsequent error calculation.

[0131] In a specific example, step S6 specifically includes:

[0132] Step S61: select a theoretical ship plate rib line, randomly select 3 points, and determine whether they are collinear. If they are collinear, reselect them. If they are not collinear, calculate the plane parameter equation Ax+By+Cz+D=0 through the selected 3 points, where A, B, C are the plane normal vector components, and D is the plane offset;

[0133] Step S62: For each point in the theoretical ship plate rib line, calculate the point (x 0 ,y 0 ,z 0 ) to the fitting plane, the calculation formula is as follows:

[0134]

[0135] If the distance is less than the set threshold, it is considered an "inside point", and if it exceeds the threshold, it is considered an "outside point";

[0136] Step S63: Count the number of all inliers, which are points that fit the current plane model. Repeat steps 61-62 multiple times, and each time randomly select 3 points from the data set to fit the plane, calculate the inliers, and record the number of inliers. If the number of inliers obtained in a certain iteration exceeds the previous maximum number of inliers, save the current plane model.

[0137] Step S64: repeating steps 61-63 until the plane fitting of each theoretical ship plate rib line is completed;

[0138] Step S65: intercepting the rib lines in the measured ship plate point cloud through these fitting planes, that is, traversing the measured ship plate point cloud, substituting its coordinates into the fitting plane equation, and if its coordinates conform to the plane equation, it is regarded as the measured ship plate rib line, so as to extract the measured ship plate rib line;

[0139] Step S66: After the plate rib line extraction is completed, the error between the two is calculated, which is the plate rib line forming deviation;

[0140] Step S67: Search the measured point cloud for the point closest to the theoretical point cloud, compare the theoretical model point cloud and the measured ship plate, calculate the Euclidean distance between the two, form a color difference cloud map, and display the overall forming deviation of the ship plate.

[0141] This embodiment innovatively uses the RANSAC plane fitting algorithm to fit the plane where the theoretical ship plate rib line is located, and uses this method to intercept the rib line point cloud on the measured ship plate, so as to calculate the rib line error and the global error.

[0142] In a specific example, the scanning device in step S1 is a 3D camera.

[0143] In the specific implementation, 9 3D cameras can be used to collect and stitch the ship plate measurement point cloud data.

[0144] Fig. 9 After the B-spline fitting interpolation of the characteristic points of the theoretical ship plate rib line, the theoretical ship plate is registered with the measured ship plate. At this time, the theoretical ship plate rib line has been aligned with the measured ship plate. The plane where each theoretical ship plate rib line is located is fitted by the RANSAC plane fitting algorithm, and the parameter equation of each plane is obtained. Then, the measured ship plate point cloud is traversed, and the coordinates of the measured ship plate point cloud are substituted into the above-mentioned fitted plane parameter equation. In this way, the measured ship plate rib line is extracted, and then the rib line error is calculated by calculating the Euclidean distance between the neighboring points of the theoretical ship plate rib line and the measured ship plate rib line. Fig. 9 It can be seen that the red rib line is the measured ship plate rib line, and the green rib line is the theoretical ship plate rib line. The rib line fitting and extraction effect is good, which lays a good foundation for the ship plate rib line error calculation.

[0145] Fig.10 The theoretical ship plate STP file is converted into a point cloud by triangular patch interpolation, and then it is registered with the measured ship plate through a registration algorithm. The nearest point to the theoretical point cloud is searched in the measured point cloud, and the point pairs of the measured ship plate point cloud and the theoretical ship plate point cloud are matched. The Euclidean distance between adjacent point pairs is calculated and the error threshold is set. The error is represented by different RGB color differences to provide a basis for subsequent production and manufacturing.

[0146] Example 2

[0147] To achieve the above purpose, see Fig.11:This embodiment also provides a water-fire bending plate forming deviation measurement system based on three-dimensional point cloud, including the following modules:

[0148] Scanning module: Use scanning equipment to collect original point cloud data, splice it through software algorithm to obtain ship plate measurement point cloud data, and obtain ship plate rib line feature point data and ship plate CAD design files through ship plate design software;

[0149] Fitting module: Use scanning equipment to collect the measured point cloud data of the ship plate, and use the ship plate design software to obtain the sparse feature point data of the ship plate rib line and the ship plate CAD design file;

[0150] Conversion module: Use the OCC geometry library to convert the shipboard CAD design file from STP format to OBJ format, and then use the triangular patch point cloud interpolation algorithm on the surface of the OBJ model file to make it a theoretical model point cloud;

[0151] Preprocessing module: used to preprocess the shipboard measurement point cloud data obtained by actual scanning using a voxel size adaptive downsampling algorithm;

[0152] Registration module: used to roughly register the shipboard measurement point cloud and the theoretical model point cloud using the improved rough registration algorithm, and then use the ICP registration algorithm for fine registration;

[0153] Forming deviation measurement module: The plane where the theoretical ship plate rib line is located is obtained by using the RANSAC plane fitting algorithm, and the rib line on the measured ship plate is intercepted by this plane, the error between the two is calculated, and the forming deviation of the ship plate rib line is determined; the nearest point to the theoretical point cloud is searched in the measured point cloud, the error between the two points of the ship plate measured point cloud and the theoretical model point cloud is calculated, and the overall forming deviation of the ship plate is displayed through the color difference cloud map.

[0154] The water-fire bending plate forming deviation measurement system based on three-dimensional point cloud of this embodiment has the same advantages as the above-mentioned water-fire bending plate forming deviation measurement method based on three-dimensional point cloud over the prior art, which will not be repeated here.

[0155] Example 3

[0156] In order to achieve the above-mentioned purpose, the present embodiment also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the water-fire bending plate forming deviation measurement method based on three-dimensional point cloud as described above is implemented.

[0157] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for measuring the forming deviation of water-fire bending plate based on three-dimensional point cloud, characterized in that: The following steps are involved: S1: Use scanning equipment to collect the measured point cloud data of the ship plate, and use the ship plate design software to obtain the sparse feature point data of the ship plate rib line and the ship plate CAD design file; S2: Use the B-spline curve fitting algorithm to perform curve fitting on the sparse feature point data of the ship plate rib line and interpolate the rib line point cloud based on the feature points; S3: Use the OCC geometry library to convert the shipboard CAD design file from STP format to OBJ format, and then use the triangular patch point cloud interpolation algorithm on the surface of the OBJ model file to make it a theoretical model point cloud; S4: Preprocessing the measured point cloud data of the shipboard obtained by actual scanning using a voxel size adaptive downsampling algorithm; S5: The improved coarse registration algorithm is used to coarsely register the shipboard measurement point cloud and the theoretical model point cloud, and then the ICP registration algorithm is used for fine registration; S6: The plane where the theoretical ship plate rib line is located is obtained by using the RANSAC plane fitting algorithm. The measured ship plate point cloud surface is intercepted by the plane to obtain the measured ship plate rib line point cloud. The error between the rib lines is calculated, which is the ship plate rib line forming deviation. The nearest point to the theoretical point cloud is searched in the measured point cloud. The error between the two points of the ship plate measured point cloud and the theoretical model point cloud is calculated. The overall forming deviation of the ship plate is displayed through the color difference cloud map.

2. The method for measuring the forming deviation of water-fire bending plate based on three-dimensional point cloud according to claim 1 is characterized in that: The step S2 specifically includes: Step S21: storing each rib line in a container according to the rib line number segment in the PCD point cloud file of the ship plate rib line feature points; Step S22: performing B-spline curve fitting on each rib feature point in the container; Step S23: performing point cloud sampling interpolation on the fitted rib curve.

3. The method for measuring the forming deviation of water-fire bending plate based on three-dimensional point cloud according to claim 1 is characterized in that: The step S3 specifically includes: Step S31: read the STP file through the OCC geometry core library and convert the STP file into OBJ format; Step S32: Use the point cloud open source library PCL to read the OBJ file and convert the data into a data format that can be processed by PCL; Step S33: traverse all the triangles in the mesh file, obtain the three vertices O, A, and B of the triangle, calculate the vectors of the two edges OA and OB, and calculate the area S of the triangle through the vectors OA and OB. The formula is: Step S34: derive the number of points to be inserted, generate a total number of points N, and insert density Den, then: N=S×Den Step S35: processing the decimal part of the generated points, and rounding the generated points to an integer using the principle of rounding off; Step S36: After obtaining the number of generated points, interpolation is performed inside the triangle patch. First, two random numbers x∈(0,1) and y∈(0,1) are generated. Then, it is determined whether x+y is greater than 1. If it is greater than 1, it means that the generated point will be outside the AB edge, that is, the generated point is not inside the triangle. Then, x and y are processed to make x=1-x, y=1-y, that is, the point cloud is generated inside the triangle. Step S37: Generate points O is a vertex of a triangle, x and y are parameters used to control the offset of P in the OA and OB directions. The generated point P is placed in the point cloud container until all triangular faces are traversed.

4. The method for measuring the deviation of water-fire plate bending based on three-dimensional point cloud according to claim 1 is characterized in that: The step S4 specifically includes: Step S41: traverse the entire point cloud and calculate the average Euclidean distance Δd between neighboring points of the entire point cloud; Step S42: The average Euclidean distance calculated in step S41 is enlarged by n times, ie, nΔd, as the input of the voxel size in the voxel filtering downsampling algorithm, so as to realize the voxel size adaptive downsampling of the measured ship plate point cloud data.

5. The method for measuring the forming deviation of water-fire bending plate based on three-dimensional point cloud according to claim 1 is characterized in that: The step S5 specifically includes: Step S51: Calculate the covariance matrix and centroid C of the pre-processed shipboard measurement point cloud data S , solve its eigenvector U through the covariance matrix S (i), where i = 1, 2, 3, which represent the main directions of the three axes x, y, and z, respectively. The new matrix U is constructed by the three eigenvectors; Step S52: Calculate the covariance matrix and centroid C of the theoretical point cloud data of the shipboard t , solve its eigenvector U through the covariance matrix t (i), where i = 1, 2, 3, a new matrix V is constructed from three eigenvectors; Step S53: Solve the rotation matrix R and translation matrix T, where: R=UV T T=C S -RC t Step S54: roughly aligning the theoretical point cloud data of the shipboard with the point cloud data of the measurement board through a rotation and translation matrix; Step S55: Calculate the RMS root mean square error E between the two point clouds after the rough registration is completed r , that is, the registration error calculation; Step S56: Transform the feature vector U S (i) Take the inverse, that is, 2 3 In each of these eight cases, the root mean square error E is discussed. r , repeat steps S4.1-S4.5 to find E r The corresponding rotation matrix R and translation matrix T when the minimum is obtained are the results of rough registration; Step S57: Finally, the measured plate point cloud and the theoretical ship plate point cloud are precisely registered by ICP based on SVD decomposition; Step S58: First, use the exhaustive search method to measure each point p in the shipboard point cloud P. i , find and p in the theoretical ship plate point cloud Q i The point q that is closest to the point i ,q i As p i The corresponding points of form a total of n corresponding point pairs; Step S59: Minimize the objective function value of the formula f(R, T), and solve the rotation matrix R and the translation matrix T through SVD decomposition; Step S510: using the obtained transformation matrix to update the source point cloud P, obtaining a new point cloud P′, and using the new point cloud as a new measurement ship plate point cloud; P′=RP+T Step S511: Repeat the process of step S57 to step S510 until the iteration condition is met, and the iteration condition is to reach the maximum number of iterations K max , or the root mean square error d between iterative corresponding points k Or the root mean square error of two iterations d k -d k-1 Less than a certain value; Step S512: The rotation and translation parameters obtained when the iteration is stopped are used as the optimal transformation matrix. According to the parameters, the measured ship plate point cloud is transformed into the theoretical ship plate point cloud, and finally the point cloud registration is completed.

6. The method for measuring the forming deviation of water-fire bending plate based on three-dimensional point cloud according to claim 1 is characterized in that: The step S6 specifically includes: Step S61: select a theoretical ship plate rib line, randomly select 3 points, and determine whether they are collinear. If they are collinear, reselect them. If they are not collinear, calculate the plane parameter equation Ax+By+Cz+D=0 through the selected 3 points, where A, B, C are the plane normal vector components, and D is the plane offset; Step S62: For each point in the theoretical ship plate rib line, calculate the distance d from the point (x0, y0, z0) to the fitting plane, and the calculation formula is as follows: If the distance is less than the set threshold, it is considered an "inside point", and if it exceeds the threshold, it is considered an "outside point"; Step S63: Count the number of all inliers, which are points that fit the current plane model. Repeat steps 61-62 multiple times, and each time randomly select 3 points from the data set to fit the plane, calculate the inliers, and record the number of inliers. If the number of inliers obtained in a certain iteration exceeds the previous maximum number of inliers, save the current plane model. Step S64: repeating steps 61 to 63 until the plane fitting of each theoretical ship plate rib line is completed; Step S65: intercepting the rib lines in the measured ship plate point cloud through these fitting planes, that is, traversing the measured ship plate point cloud, substituting its coordinates into the fitting plane equation, and if its coordinates conform to the plane equation, it is regarded as the measured ship plate rib line, so as to extract the measured ship plate rib line; Step S66: After the plate rib line extraction is completed, the error between the two is calculated, which is the plate rib line forming deviation; Step S67: Search the measured point cloud for the point closest to the theoretical point cloud, compare the theoretical model point cloud and the measured ship plate point cloud, calculate the Euclidean distance between the two, form a color difference cloud map, and display the overall forming deviation of the ship plate.

7. The method for measuring the forming deviation of water-fire bending plate based on three-dimensional point cloud according to claim 1 is characterized in that: The scanning device in step S1 is a 3D camera.

8. A water-fire bending plate forming deviation measurement system based on three-dimensional point cloud, characterized in that: Includes the following modules: Scanning module: Use scanning equipment to collect the measured point cloud data of the ship plate, and use the ship plate design software to obtain the sparse feature point data of the ship plate rib line and the ship plate CAD design file; Fitting module: Use B-spline curve fitting algorithm to perform curve fitting on the sparse feature point data of the ship plate rib line and interpolate the rib line point cloud based on the feature points; Conversion module: Use the OCC geometry library to convert the shipboard CAD design file from STP format to OBJ format, and then use the triangular patch point cloud interpolation algorithm on the surface of the OBJ model file to make it a theoretical model point cloud; Preprocessing module: used to preprocess the shipboard measurement point cloud data obtained by actual scanning using a voxel size adaptive downsampling algorithm; Registration module: used to roughly register the shipboard measurement point cloud and the theoretical model point cloud using the improved rough registration algorithm, and then use the ICP registration algorithm for fine registration; Forming deviation measurement module: The plane where the theoretical ship plate rib line is located is obtained by using the RANSAC plane fitting algorithm, and the rib line on the measured ship plate is intercepted by this plane, the error between the two is calculated, and the forming deviation of the ship plate rib line is determined; the nearest point to the theoretical point cloud is searched in the measured point cloud, the error between the two points of the ship plate measured point cloud and the theoretical model point cloud is calculated, and the overall forming deviation of the ship plate is displayed through the color difference cloud map.

9. A computer device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for measuring the forming deviation of a water-fire bending plate based on a three-dimensional point cloud as described in any one of claims 1 to 7 is implemented.

Images