A method and device for removing a three-dimensional point cloud and a detection system

Abstract

The embodiment of the application discloses a kind of rejection methods, device and detection system of three-dimensional point cloud;The method can include: according to linear acquisition three-dimensional point cloud data, according to the plane in which the surface of processing workpiece top view is located and linear acquisition direction vertical direction is carried out data slice, obtain multiple slice data groups;For each slice data group, construct segmentation standard vector, and obtain the initial screening air film hole hole wall point cloud data from each slice data group according to the segmentation standard vector;According to the initial screening hole wall point cloud data, cylindrical fitting is carried out, and the initial axis of air film hole is obtained;According to the initial axis, rotation matrix is calculated;The initial screening hole wall point cloud data is carried out coordinate transformation using the rotation matrix, and air film hole orifice point cloud data is screened;From the initial screening hole wall point cloud data, the point cloud data within the set distance threshold with the air film hole orifice point cloud data is rejected, to obtain the final air film hole hole wall point cloud data.

Description

Technical Field

[0001] The embodiments of the present invention relate to micro-hole processing and inspection technology, and more particularly to a method, apparatus and inspection system for removing three-dimensional point clouds. Background Technology

[0002] In precision machining, it is necessary to machine multiple micro-holes on the surface of a three-dimensional curved workpiece. For example, consider a turbine blade in an aero-engine. Figure 1 As shown, in order to improve reliability under high mechanical load conditions, multiple film cooling holes are usually machined on the surface of turbine blades. The diameter of the film cooling holes is mostly within 1 mm, and the axial direction, position distribution, geometry and other aspects have strict precision requirements.

[0003] To meet accuracy requirements, inspection is typically performed by acquiring three-dimensional point cloud data of the blade. Understandably, the acquired three-dimensional point cloud includes point cloud data of the micropore or film pore wall region as well as point cloud data of the blade surface region. During the inspection process using point cloud data, the point cloud data of the blade surface region needs to be removed from the acquired point cloud data. Only the point cloud data of the pore wall region obtained after removal is used to inspect the processing accuracy of the micropore or film pore. In other words, the accuracy of acquiring the point cloud data of the micropore or film pore directly determines the accuracy of the inspection of the film pore.

[0004] Based on this, a commonly used point cloud removal scheme extracts the plane containing the blade surface from the highest Z-axis coordinate in the point cloud, then removes all point clouds on that plane. The remaining point clouds are the hole wall point clouds for the film air vents. It's worth noting that blade surfaces are usually not planar but curved. Therefore, this scheme is unsuitable for blades with large surface curvatures, as it would remove a large amount of hole wall point cloud data during implementation, thus reducing the removal accuracy. Another commonly used point cloud removal scheme uses the Random Sample Consensus (RANSAC) algorithm to remove all points on the processing plane before performing cylindrical fitting. However, because this algorithm has low requirements for the initial data processing, the final fitted hole diameter error is relatively large. Summary of the Invention

[0005] In view of this, the present invention aims to provide a method, apparatus and detection system for removing three-dimensional point clouds; it can improve the accuracy of removing point cloud data when the surface curvature of the processed workpiece is large, thereby improving the accuracy of fitting the cylinder of the air film hole in the subsequent detection process.

[0006] The technical solution of this invention is implemented as follows:

[0007] In a first aspect, embodiments of the present invention provide a method for removing points from a three-dimensional point cloud, the method comprising:

[0008] Based on the linearly acquired 3D point cloud data, data slices are made in the plane of the workpiece surface top view, perpendicular to the linear acquisition direction, to obtain multiple slice data groups.

[0009] For each slice data group, a segmentation standard vector is constructed, and the initial screening point cloud data of the air film pore wall is obtained from each slice data group based on the segmentation standard vector.

[0010] Based on the point cloud data of the pore wall after the initial screening, a cylindrical fitting is performed to obtain the initial axis of the air film pores;

[0011] A rotation matrix is ​​calculated based on the initial axis, wherein the rotation matrix represents rotating the initial axis so that it is parallel to the vertical axis of the plane containing the top view of the workpiece surface;

[0012] The rotation matrix is ​​used to perform coordinate transformation on the point cloud data of the hole wall after the initial screening, and the point cloud data of the air film pore opening is then screened.

[0013] Point cloud data that are within a set distance threshold from the point cloud data of the air film pore opening are removed from the initial screened pore wall point cloud data to obtain the final air film pore wall point cloud data.

[0014] Secondly, embodiments of the present invention provide a three-dimensional point cloud rejection device, the device comprising: a slicing section, a preliminary screening section, a fitting section, a rotation section, a coordinate transformation section, and a rejection section; wherein...

[0015] The slicing portion is configured to slice the data based on the linearly acquired three-dimensional point cloud data, according to the plane where the top view of the surface of the processed workpiece is located, in a direction perpendicular to the linear acquisition direction, to obtain multiple slice data groups.

[0016] The initial screening section is configured to construct a segmentation standard vector for each slice data group, and to filter out the initial screening air film pore wall point cloud data from each slice data group according to the segmentation standard vector.

[0017] The fitting part is configured to perform cylindrical fitting based on the point cloud data of the pore wall after the initial screening to obtain the initial axis of the air film pores.

[0018] The rotating portion is configured to calculate a rotation matrix based on the initial axis, wherein the rotation matrix represents rotating the initial axis so that it is parallel to the vertical axis of the plane containing the top view of the workpiece surface;

[0019] The coordinate transformation part is configured to use the rotation matrix to perform coordinate transformation on the point cloud data of the hole wall after the initial screening, and to screen the point cloud data of the air film hole opening.

[0020] The rejection section is configured to remove point cloud data that are within a set distance threshold from the point cloud data of the air film pore opening after the initial screening, so as to obtain the final point cloud data of the air film pore wall.

[0021] Thirdly, embodiments of the present invention provide a detection system, the detection system comprising: a numerical control device, a motion platform, a point cloud data scanning device, and a three-dimensional point cloud removal device as described in the second aspect; wherein...

[0022] The motion platform is connected to the point cloud data scanning device and the CNC device, and controls the point cloud data scanning device to linearly acquire three-dimensional laser point cloud data of the workpiece surface. The three-dimensional point cloud removal device is connected to the point cloud data scanning device.

[0023] Fourthly, embodiments of the present invention provide a computer storage medium storing a culling program for a three-dimensional point cloud, wherein when the culling program for the three-dimensional point cloud is executed by at least one processor, it implements the steps of the culling method for the three-dimensional point cloud described in the first aspect.

[0024] This invention provides a method, apparatus, and detection system for removing point clouds in three-dimensional images. First, point cloud data of the air film hole walls are initially screened. An initial axis is obtained by combining this with cylindrical fitting. Then, all the initially screened air film hole wall point cloud data are rotated to the plane of the workpiece surface top view based on the initial axis. The point cloud data of the air film hole openings and the noise points along the edges of the openings are then removed, resulting in the final air film hole wall point cloud data. This method can remove noise points along the edges of the air film hole openings when the workpiece surface has a large curvature, improving the accuracy of point cloud data removal and enhancing the effectiveness of the air film hole wall data. This, in turn, improves the accuracy of cylindrical fitting of the air film holes during subsequent detection. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of a turbine blade;

[0026] Figure 2 A schematic flowchart of a three-dimensional point cloud removal method provided in an embodiment of the present invention;

[0027] Figure 3 This is a schematic diagram of point cloud data of air film pores provided in an embodiment of the present invention;

[0028] Figure 4 This is a schematic top view of the surface of the workpiece provided in an embodiment of the present invention;

[0029] Figure 5 A vector diagram provided for an embodiment of the present invention;

[0030] Figure 6 This is a schematic diagram of the point cloud data of the air film pore wall after initial screening, provided in an embodiment of the present invention.

[0031] Figure 7 A schematic diagram of the geometric characteristics of a cylinder provided in an embodiment of the present invention;

[0032] Figure 8 This is a schematic diagram of the initial axis of the air film pore provided in an embodiment of the present invention;

[0033] Figure 9 This is a schematic diagram of the transformed hole wall point cloud data provided in an embodiment of the present invention;

[0034] Figure 10 A schematic diagram of the final hole wall point cloud data provided for an embodiment of the present invention;

[0035] Figure 11 A schematic diagram of a three-dimensional point cloud removal device provided in an embodiment of the present invention;

[0036] Figure 12 This is a schematic diagram of the composition structure of a detection system provided in an embodiment of the present invention. Detailed Implementation

[0037] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0038] To clearly illustrate the technical solutions of this disclosure, the embodiments of this invention are illustrated using air film holes as an example. It is understood that those skilled in the art can apply the technical solutions of the embodiments of this invention to the processing of micro-holes with a diameter of 0.3-0.5 mm. The embodiments of this invention will not be described in detail here.

[0039] See Figure 2 This illustrates a method for removing three-dimensional point clouds according to an embodiment of the present invention, the method comprising:

[0040] S201: Based on the linearly acquired 3D point cloud data, data slices are made in the plane of the workpiece surface top view, perpendicular to the linear acquisition direction, to obtain multiple slice data groups.

[0041] S202: For each slice data group, construct a segmentation standard vector, and select the initial screening air film pore wall point cloud data from each slice data group according to the segmentation standard vector;

[0042] S203: Perform cylindrical fitting based on the point cloud data of the pore wall after the initial screening to obtain the initial axis of the air film pore;

[0043] S204: Calculate a rotation matrix based on the initial axis, wherein the rotation matrix represents rotating the initial axis so that it is parallel to the vertical axis of the plane containing the top view of the workpiece surface;

[0044] S205: Use the rotation matrix to perform coordinate transformation on the point cloud data of the hole wall after the initial screening, and then screen the point cloud data of the air film pore opening.

[0045] S206: Remove point cloud data that are within a set distance threshold from the point cloud data of the air film pore opening after the initial screening from the point cloud data of the pore wall, so as to obtain the final point cloud data of the air film pore wall.

[0046] pass Figure 2 The technical solution described above first screens the point cloud data of the film-forming hole walls, then uses cylindrical fitting to obtain an initial axis. Next, all the pre-screened point cloud data of the film-forming hole walls are rotated to the plane of the workpiece surface top view based on the initial axis. The point cloud data of the film-forming hole openings and noise points along the opening edges are then removed, resulting in the final point cloud data of the film-forming hole walls. This solution can remove noise points along the opening edges of film-forming holes when the workpiece surface has a large curvature, improving the accuracy of the removed point cloud data and enhancing the effectiveness of the film-forming hole wall data. This, in turn, improves the accuracy of cylindrical fitting of the film-forming holes during subsequent inspection.

[0047] for Figure 2 In some implementations of the technical solution shown, the step of slicing the linearly acquired 3D point cloud data into multiple slice data groups based on the plane containing the top view of the workpiece surface, perpendicular to the linear acquisition direction, includes:

[0048] A planar coordinate system is constructed using a top view of the workpiece surface. The linear acquisition direction of the point cloud data is taken as the positive X-axis. The acquired point cloud data is sliced ​​according to the Y-axis coordinate values ​​to obtain multiple slice data groups.

[0049] Specifically, regarding the above implementation method, Figure 1 Taking any one of the air film pores as an example, its point cloud data is as follows: Figure 3 As shown, in Figure 3 In the diagram, the surface area of ​​the machined workpiece (e.g., a blade) is labeled as region A, and the wall of the film air hole is labeled as region B. A Keyence measuring head is used to acquire point cloud data linearly, with the film air hole center as the intersection point for field-of-view image acquisition. The top view of the machined workpiece surface is shown below. Figure 4As shown in the diagram. Based on this top view, a planar coordinate system is constructed, with the linear acquisition direction as the positive X-axis and the axis perpendicular to the planar coordinate system as the Z-axis, thus constructing a spatial coordinate system. Each point cloud data corresponds to a point in the coordinate system. Then, the acquired point cloud data is sliced ​​according to the Y-axis to obtain n slice data groups. It can be understood that the Y-axis coordinate values ​​of the point cloud data in each slice data group are consistent.

[0050] for Figure 2 In some implementations of the technical solution shown, the step of constructing a segmentation standard vector for each slice data group and obtaining the initial screened air film pore wall point cloud data from each slice data group based on the segmentation standard vector includes:

[0051] For each slice of data group, perform the following steps:

[0052] In the sliced ​​data group, a segmentation standard vector is constructed based on the coordinates of a set number of point cloud data at the start and end positions along the linear acquisition direction.

[0053] Establish a corresponding planar vector between the coordinates of the other point cloud data in the slice data group, excluding the set number of point cloud data at the start and end positions, and one of the farthest vertices of the segmentation standard vector;

[0054] The point cloud data of the workpiece surface is determined according to the angle between the plane vector and the segmentation standard vector;

[0055] The vector to be evaluated is constructed based on the plane vectors corresponding to the remaining point cloud data (excluding the point cloud data of the workpiece surface) in the other point cloud data and the segmentation standard vector.

[0056] The point cloud data of the air film pore walls after initial screening is obtained by rotating the vector to be evaluated to the rotation axis of the segmentation standard vector.

[0057] For the above implementation method, combined with the specific example mentioned above, two queues can be set up for each slice data group to store the point cloud of region A and the point cloud of region B respectively. In the specific implementation process, the queue storing the point cloud data of region A and the queue storing the point cloud data of region B are named QueSurface and QueHole respectively.

[0058] Next, for each slice of data group, the following processing procedure is performed:

[0059] Ten coordinate points at the beginning and ten coordinate points at the end of each group are selected along the linear acquisition direction. The mean coordinates of these ten points are calculated using the mean method, and are defined as the mean point S(x,y,z) at the beginning of the group and the mean point E(x,y,z) at the end of the group, respectively. Figure 5 As shown, a standard segmentation vector V0(x,y,z) = E(x,y,z) - S(x,y,z) is constructed from point S to point E. It can be understood that since the Y-axis coordinate values ​​of each slice data group are equal, therefore... Figure 5 The vectors shown are all located in the XOZ coordinate system of the spatial coordinate system obtained above.

[0060] For the other point cloud data coordinates P(x,y,z) in each slice data group, such as Figure 5 The examples P1, P2, and P3 shown here establish corresponding planar vectors based on points whose coordinates are farthest from points S or E in the segmentation standard vector. For example, if D... P-S >D P-E If the condition is met, then establish the corresponding plane vector V(x,y,z) = P(x,y,z) - S(x,y,z) for point P; otherwise, establish the corresponding plane vector V(x,y,z) = E(x,y,z) - P(x,y,z) for point P.

[0061] First, based on the curvature of the workpiece surface, set the included angle threshold AngleThold, which can usually be preferably set to 15°; then, calculate the angle between the plane vector V(x,y,z) corresponding to point P and the segmentation standard vector V0(x,y,z). If the included angle is less than the included angle threshold AngleThold, it can be determined that point P is on the surface of the workpiece (such as a blade). Therefore, the point cloud data corresponding to point P is placed in the queue QueSurface.

[0062] Otherwise, construct the vector to be evaluated between point P and the starting point of the corresponding plane vector. For example, if the starting point of the plane vector corresponding to point P is set to point S in the standard vector of the segmentation, then the vector to be evaluated corresponding to point P is V. t (x,y,z)=P(x,y,z)-S(x,y,z);

[0063] With V t The rotation axis V is calculated from (x,y,z) to V0(x,y,z) as the rotation direction. r (x,y,z)=V t ×V0, and according to Figure 5 The right-hand screw rule shown in the upper right corner determines the axis of rotation V. r Is the y-component positive? Figure 5As shown, the rotation direction of the vectors V1 and V3 to be evaluated corresponding to P1 and P3 is clockwise, and the rotation direction of the vector V2 to be evaluated corresponding to P2 is counterclockwise. According to the right-hand screw rule, the direction in which the four fingers are bent is the rotation direction, so the direction pointed to by the thumb represents the rotation axis V. r The direction of the y-component, specifically, if the rotation axis V r If the direction of the y-component is positive, then... Figure 5 The direction inward is indicated by ×; if the axis of rotation V r If the direction of the y-component is negative, then... Figure 5 To indicate an outward direction, use a dot. ● express.

[0064] In the specific implementation process, positive point cloud data and negative point cloud data are stored separately. For example, positive point cloud data is stored in a queue named QueRight, and negative point cloud data is stored in a queue named QueLeft. Then, the point cloud data of the queue with more point cloud data in the QueRight and QueLeft queues is used as the point cloud data of the air film pore wall after the initial screening of the slice data group.

[0065] It should be noted that the segmentation standard vector can be considered as the surface of the machined workpiece, and used as... Figure 5 The X-axis in the XOZ coordinate system is shown, and the Z-axis represents the height from the surface of the workpiece. It can be understood that for film hole machining, the number of point cloud data for the film hole wall is usually significantly greater than the number of point cloud data for the surface region A of the workpiece. Therefore, for film hole machining on the workpiece surface from different directions, it is necessary to count the point cloud data in both rotation directions to identify the larger number as the film hole wall. This is the reason why the statistics are counted according to the positive and negative y-components of the rotation axis in the aforementioned specific example.

[0066] Through the above implementation methods and specific examples, Figure 3 The point cloud data of the pore walls of the air film pores obtained after initial screening, as shown in the example of the air film pores, are as follows: Figure 6 As shown in the mid-highlighted portion, it can be seen that in Figure 6 In the middle, the point cloud of the air film pore wall is located below the surface.

[0067] for Figure 2 In some implementations of the technical solution shown, the step of performing cylindrical fitting based on the point cloud data of the pore walls after initial screening to obtain the initial axis of the air film pores includes:

[0068] Set the parameters of the cylinder model;

[0069] The initial values ​​of the cylindrical model parameters are established based on the coordinates of the point cloud data of the hole wall after the initial screening.

[0070] An error equation based on model parameters is established based on the cylindrical model equation.

[0071] Based on the initial values ​​of the cylindrical model parameters and the error equation, the estimated values ​​of the cylindrical model parameters that meet the set accuracy requirements are obtained iteratively using the least squares principle.

[0072] The initial axis of the air film aperture is determined from the estimated values ​​of the cylindrical model parameters.

[0073] Regarding the above implementation, this embodiment of the invention aims to obtain the initial axis of the film pores by performing cylindrical fitting using the least squares method combined with Newton's iteration method; that is, by using the least squares method and Newton's iteration method to perform cylindrical fitting on the coordinates of the pore wall point cloud data after initial screening, the pore diameter and pore axis information of the film pores based on the pore wall point cloud data after initial screening are obtained. Specifically, by Figure 7 The geometric properties of the cylinder shown indicate that the distance from any point on the cylinder's side surface to its axis is always equal to the radius r. Here, (x, y, z) represents any point on the cylinder's surface, (x0, y0, z0) represents a point on the cylinder's axis, (a, b, c) represents the cylinder's axis vector, and r is the radius of the cylinder's base circle. Since the distance from any point on the cylinder's surface to its axis is always the radius r of the cylinder's base, using the coordinates of the point cloud data from the hole wall after initial screening as points on the cylinder to be fitted, the above seven parameters can be obtained through a fitting algorithm, thus uniquely determining a cylinder.

[0074] The specific cylinder fitting process mainly includes two steps: first, determining the initial values ​​of the cylindrical model parameters; and second, establishing error equations to solve for the parameter values. Specifically, firstly, using the least squares method, the initial values ​​of seven cylindrical model parameters are determined through the coordinates of the hole wall point cloud data after initial screening: the initial axis vector (a, b, c), a point on the cylinder axis (x0, y0, z0), and the radius r of the cylinder's base circle.

[0075] Next, an improved error equation is established, and the parameters are solved. Specifically, the equation for the cylinder is shown below:

[0076] (x-x0) 2 +(y-y0) 2 +(z-z0) 2 -[a(x-x0)+b(y-y0)+c(z-z0)] 2 =r 2

[0077] Let f = (x - x0) 2 +(y-y0) 2 +(z-z0) 2 -[a(x-x0)+b(y-y0)+c(z-z0)]2 -r 2 By linearizing f using the first-order differential, we can obtain...

[0078] Based on the above linearized expression, the error equation is obtained as follows:

[0079]

[0080] Where, f0 = (x - x0) 2 +(y-y0) 2 +(z-z0) 2 -[a(x-x0)+b(y-y0)+c(z-z0)] 2 -r 2 ;

[0081] Let f represent the expressions obtained by differentiating f with respect to parameters x0, a, r, y0, z0, b, and c, respectively; for example... The expressions obtained after differentiating other parameters will not be elaborated here.

[0082] The error equation shown above can be written in matrix form V n×1 =B n×7 X 7×1 -L n×1 X 7×1 This represents a vector composed of unknown parameters. Based on the matrix concept of the least squares method, as long as V is minimized, close to 0, that is, the points on the acquired cylindrical surface are set to have the same precision, according to the least squares principle, V... T If V = min, the unknown parameter X = (B) can be obtained iteratively. T B) - 1 B T L. In this iterative process, the initial value substituted in during each iteration is equal to the previous initial value plus the correction value of X obtained. When the value of X is small enough to meet the required accuracy, the iteration can be considered complete and the iteration can be exited, thus enabling the fitting of the initial axis of the air film pore, as follows: Figure 8 The bright lines in the image are shown.

[0083] for Figure 2 In some implementations of the technical solution shown, the step of using the rotation matrix to perform coordinate transformation on the initially screened pore wall point cloud data and then screening the air film pore opening point cloud data includes:

[0084] The hole wall point cloud data after initial screening is transformed according to the rotation matrix so that the coordinates of the transformed hole wall point cloud data are perpendicular to the plane of the top view of the workpiece surface.

[0085] Using the initial axis as the center, the circumference angle is divided into n equal parts. Among the transformed hole wall point cloud data corresponding to each equal angle, the hole wall point cloud data with the largest Z-axis coordinate value perpendicular to the plane of the top view of the workpiece surface is selected as the air film hole orifice point cloud data.

[0086] Specifically, the initial axis V(x,y,z) obtained above is rotated to be parallel to the coordinate axis Z. The rotation matrix R can be calculated using the initial axis vector and the vector parallel to the Z-axis after rotation. Then, the coordinates of the hole wall point cloud data after initial screening are transformed using this rotation matrix so that the hole wall is perpendicular to the XOY plane, as follows. Figure 9 As shown in the figure, the bright line represents the Z-axis. In the transformed hole wall point cloud data coordinates, with the axis as the center, the circumference angle of 360 degrees is divided into equal parts at 1 degree intervals. For each equal part, the coordinate point P(x,y,z) with the largest Z-axis coordinate value at X and Y points is obtained for the hole wall part, thus obtaining a maximum of 360 points. These points can be regarded as the point cloud data of the hole opening. In this embodiment of the invention, these points can be stored in a queue named ZMaxQue for use in subsequent processing.

[0087] for Figure 2 In step S206 of the technical solution shown, specifically, a threshold m (preferably set to 10µm in this embodiment) can be set. Using each point cloud data ZMaxQue[i] in the queue ZMaxQue as a reference, all point cloud data within m of this point are removed, i.e., all point cloud data within m of the hole opening are removed. The remaining point cloud data is the final hole wall point cloud data, such as... Figure 10 As shown

[0088] After obtaining the final point cloud data of the air film pore walls, using this final point cloud data to perform cylindrical fitting of the air film pores and execute subsequent detection processes can effectively improve the fitting accuracy.

[0089] During implementation, the above technical solution eliminates the influence of surface curvature and orifice point cloud quality on the air film pore surface, resulting in a 30% improvement in the accuracy of air film pore fitting using the final air film pore wall point cloud data compared to the conventional RANSAC algorithm.

[0090] Based on the same inventive concept as the aforementioned technical solution, see [link to inventive concept]. Figure 11 This illustration shows a three-dimensional point cloud rejection device 110 provided in an embodiment of the present invention. The device 110 includes: a slicing portion 1101, a preliminary screening portion 1102, a fitting portion 1103, a rotation portion 1104, a coordinate transformation portion 1105, and a rejection portion 1106; wherein,

[0091] The slicing portion 1101 is configured to slice the data according to the linearly acquired three-dimensional point cloud data, in the plane where the top view of the surface of the workpiece is located, in the direction perpendicular to the linear acquisition direction, to obtain multiple slice data groups.

[0092] The initial screening section 1102 is configured to construct a segmentation standard vector for each slice data group, and to obtain the initial screening air film pore wall point cloud data from each slice data group according to the segmentation standard vector.

[0093] The fitting part 1103 is configured to perform cylindrical fitting based on the point cloud data of the pore wall after the initial screening to obtain the initial axis of the air film pore.

[0094] The rotating portion 1104 is configured to calculate a rotation matrix based on the initial axis, wherein the rotation matrix represents rotating the initial axis so that it is parallel to the vertical axis of the plane containing the top view of the surface of the workpiece.

[0095] The coordinate transformation part 1105 is configured to use the rotation matrix to perform coordinate transformation on the point cloud data of the hole wall after the initial screening, and to screen the point cloud data of the air film hole opening.

[0096] The rejection section 1106 is configured to remove point cloud data that are within a set distance threshold from the point cloud data of the air film pore opening after the initial screening, so as to obtain the final point cloud data of the air film pore wall.

[0097] In some examples, the slice portion 1101 is configured as follows:

[0098] A planar coordinate system is constructed using a top view of the workpiece surface. The linear acquisition direction of the point cloud data is taken as the positive X-axis. The acquired point cloud data is sliced ​​according to the Y-axis coordinate values ​​to obtain multiple slice data groups.

[0099] In some examples, the initial screening section 1102 is configured as follows:

[0100] For each slice of data group, perform the following steps:

[0101] In the sliced ​​data group, a segmentation standard vector is constructed based on the coordinates of a set number of point cloud data at the start and end positions along the linear acquisition direction.

[0102] Establish a corresponding planar vector between the coordinates of the other point cloud data in the slice data group, excluding the set number of point cloud data at the start and end positions, and one of the farthest vertices of the segmentation standard vector;

[0103] The point cloud data of the workpiece surface is determined according to the angle between the plane vector and the segmentation standard vector;

[0104] The vector to be evaluated is constructed based on the plane vectors corresponding to the remaining point cloud data (excluding the point cloud data of the workpiece surface) in the other point cloud data and the segmentation standard vector.

[0105] The point cloud data of the air film pore walls after initial screening is obtained by rotating the vector to be evaluated to the rotation axis of the segmentation standard vector.

[0106] In some examples, the fitting portion 1103 is configured as follows:

[0107] Set the parameters of the cylinder model;

[0108] The initial values ​​of the cylindrical model parameters are established based on the coordinates of the point cloud data of the hole wall after the initial screening.

[0109] An error equation based on model parameters is established based on the cylindrical model equation.

[0110] Based on the initial values ​​of the cylindrical model parameters and the error equation, the estimated values ​​of the cylindrical model parameters that meet the set accuracy requirements are obtained iteratively using the least squares principle.

[0111] The initial axis of the air film aperture is determined from the estimated values ​​of the cylindrical model parameters.

[0112] In some examples, the coordinate transformation portion 1105 is configured as follows:

[0113] The hole wall point cloud data after initial screening is transformed according to the rotation matrix so that the coordinates of the transformed hole wall point cloud data are perpendicular to the plane of the top view of the workpiece surface.

[0114] Using the initial axis as the center, the circumference angle is divided into n equal parts. Among the transformed hole wall point cloud data corresponding to each equal angle, the hole wall point cloud data with the largest Z-axis coordinate value perpendicular to the plane of the top view of the workpiece surface is selected as the air film hole orifice point cloud data.

[0115] Understandably, in this embodiment, "part" can be a part of a circuit, a part of a processor, a part of a program or software, etc., or it can be a unit, a module, or a non-modular one.

[0116] Furthermore, in this embodiment, the components can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional module.

[0117] If the integrated unit is implemented as a software functional module and not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the method described in this embodiment. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0118] Therefore, this embodiment provides a computer storage medium storing a 3D point cloud culling program, which, when executed by at least one processor, implements the steps of the 3D point cloud culling method described in the above technical solution.

[0119] Based on the aforementioned 3D point cloud removal device 110 and computer storage medium, see [reference needed]. Figure 12 This illustration shows a detection system 120 provided by an embodiment of the present invention. The detection system 120 includes: a numerical control device 1201, a motion platform 1202, a point cloud data scanning device 1203, and a three-dimensional point cloud removal device 110 as described in the aforementioned technical solution; wherein,

[0120] The motion platform 1202 is connected to the point cloud data scanning device 1203 and the CNC device 1201 and controls the point cloud data scanning device 1203 to linearly acquire three-dimensional laser point cloud data of the workpiece surface. The three-dimensional point cloud rejection device 110 is connected to the point cloud data scanning device 1203.

[0121] Specifically, the point cloud data scanning device 1203 is placed on the motion platform 1202, the workpiece is set on the fixed platform, and the CNC device 1201 controls the motion direction and speed of the motion platform 1202, so that the point cloud data scanning device 1203 can linearly collect the three-dimensional laser point cloud data of the workpiece surface and transmit the collected three-dimensional laser point cloud data of the workpiece surface to the three-dimensional point cloud rejection device 110.

[0122] In some examples, the point cloud data scanning device 1203 is preferably a Keyence measuring head.

[0123] In some examples, the three-dimensional point cloud removal device 110 can obtain the final air film pore wall point cloud data by executing the three-dimensional point cloud removal method described in the aforementioned technical solution, which will not be elaborated further in this embodiment of the invention.

[0124] It is understood that the exemplary technical solutions of the above-described 3D point cloud removal device 110 and detection system 120 belong to the same concept as the aforementioned 3D point cloud removal method. Therefore, all details not described in detail above regarding the technical solutions of the 3D point cloud removal device 110 and detection system 120 can be found in the description of the aforementioned 3D point cloud removal method. This embodiment of the invention will not elaborate further on these details.

[0125] It should be noted that the technical solutions described in the embodiments of the present invention can be combined arbitrarily without conflict.

[0126] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for removing points from a three-dimensional point cloud, characterized in that, The method includes: Based on the linearly acquired 3D point cloud data, data slices are made in the plane of the workpiece surface top view, perpendicular to the linear acquisition direction, to obtain multiple slice data groups. For each slice data group, a segmentation standard vector is constructed, and the initial screening point cloud data of the air film pore wall is obtained from each slice data group based on the segmentation standard vector. Based on the point cloud data of the pore wall after the initial screening, a cylindrical fitting is performed to obtain the initial axis of the air film pores; A rotation matrix is ​​calculated based on the initial axis, wherein the rotation matrix represents rotating the initial axis so that it is parallel to the vertical axis of the plane containing the top view of the workpiece surface; The rotation matrix is ​​used to perform coordinate transformation on the point cloud data of the hole wall after the initial screening, and the point cloud data of the air film pore opening is then screened. Point cloud data that are within a set distance threshold from the point cloud data of the air film pore opening are removed from the initial screened pore wall point cloud data to obtain the final air film pore wall point cloud data. The step of constructing a segmentation standard vector for each slice data group and obtaining the initial screening point cloud data of the air film pore walls from each slice data group based on the segmentation standard vector includes: For each slice of data group, perform the following steps: In the sliced ​​data group, a segmentation standard vector is constructed based on the coordinates of a set number of point cloud data at the start and end positions along the linear acquisition direction. Establish a corresponding planar vector between the coordinates of the other point cloud data in the slice data group, excluding the set number of point cloud data at the start and end positions, and one of the farthest vertices of the segmentation standard vector; The point cloud data of the workpiece surface is determined according to the angle between the plane vector and the segmentation standard vector; The vector to be evaluated is constructed based on the plane vectors corresponding to the remaining point cloud data (excluding the point cloud data of the workpiece surface) in the other point cloud data and the segmentation standard vector. The point cloud data of the air film pore walls after initial screening is obtained by rotating the vector to be evaluated to the rotation axis of the segmentation standard vector.

2. The method according to claim 1, characterized in that, The process involves slicing the linearly acquired 3D point cloud data into multiple slice data groups based on the plane containing the top view of the workpiece surface, perpendicular to the linear acquisition direction. A planar coordinate system is constructed using a top view of the workpiece surface. The linear acquisition direction of the point cloud data is taken as the positive X-axis. The acquired point cloud data is sliced ​​according to the Y-axis coordinate values ​​to obtain multiple slice data groups.

3. The method according to claim 1, characterized in that, The step of performing cylindrical fitting based on the point cloud data of the pore walls after initial screening to obtain the initial axis of the air film pores includes: Set the parameters of the cylinder model; The initial values ​​of the cylindrical model parameters are established based on the coordinates of the point cloud data of the hole wall after the initial screening. An error equation based on model parameters is established based on the cylindrical model equation. Based on the initial values ​​of the cylindrical model parameters and the error equation, the estimated values ​​of the cylindrical model parameters that meet the set accuracy requirements are obtained iteratively using the least squares principle. The initial axis of the air film aperture is determined from the estimated values ​​of the cylindrical model parameters.

4. The method according to claim 1, characterized in that, The step of using the rotation matrix to perform coordinate transformation on the initially screened pore wall point cloud data and then screening the air film pore opening point cloud data includes: The hole wall point cloud data after initial screening is transformed according to the rotation matrix so that the coordinates of the transformed hole wall point cloud data are perpendicular to the plane of the top view of the workpiece surface. Using the initial axis as the center, the circumference angle is divided into n equal parts. Among the transformed hole wall point cloud data corresponding to each equal angle, the hole wall point cloud data with the largest Z-axis coordinate value perpendicular to the plane of the top view of the workpiece surface is selected as the air film hole orifice point cloud data.

5. A three-dimensional point cloud removal device, characterized in that, The device includes: a slicing section, a preliminary screening section, a fitting section, a rotation section, a coordinate transformation section, and a rejection section; wherein... The slicing portion is configured to slice the data based on the linearly acquired three-dimensional point cloud data, according to the plane where the top view of the surface of the processed workpiece is located, in a direction perpendicular to the linear acquisition direction, to obtain multiple slice data groups. The initial screening section is configured to construct a segmentation standard vector for each slice data group, and to filter out the initial screening air film pore wall point cloud data from each slice data group according to the segmentation standard vector. The fitting part is configured to perform cylindrical fitting based on the point cloud data of the pore wall after the initial screening to obtain the initial axis of the air film pores. The rotating portion is configured to calculate a rotation matrix based on the initial axis, wherein the rotation matrix represents rotating the initial axis so that it is parallel to the vertical axis of the plane containing the top view of the workpiece surface; The coordinate transformation section is configured to use the rotation matrix to perform coordinate transformation on the point cloud data of the hole wall after the initial screening, and to screen the point cloud data of the air film pore opening. The rejection section is configured to remove point cloud data that are within a set distance threshold from the point cloud data of the air film pore opening after the initial screening, so as to obtain the final point cloud data of the air film pore wall. The initial screening process is configured to construct a segmentation standard vector for each slice data group, and to filter the initial screening point cloud data of the air film pore walls from each slice data group based on the segmentation standard vector, including: For each slice of data group, perform the following steps: In the sliced ​​data group, a segmentation standard vector is constructed based on the coordinates of a set number of point cloud data at the start and end positions along the linear acquisition direction. Establish a corresponding planar vector between the coordinates of the other point cloud data in the slice data group, excluding the set number of point cloud data at the start and end positions, and one of the farthest vertices of the segmentation standard vector; The point cloud data of the workpiece surface is determined according to the angle between the plane vector and the segmentation standard vector; The vector to be evaluated is constructed based on the plane vectors corresponding to the remaining point cloud data (excluding the point cloud data of the workpiece surface) in the other point cloud data and the segmentation standard vector. The point cloud data of the air film pore walls after initial screening is obtained by rotating the vector to be evaluated to the rotation axis of the segmentation standard vector.

6. The apparatus according to claim 5, characterized in that, The fitting portion is configured as follows: Set the parameters of the cylinder model; The initial values ​​of the cylindrical model parameters are established based on the coordinates of the point cloud data of the hole wall after the initial screening. An error equation based on model parameters is established based on the cylindrical model equation. Based on the initial values ​​of the cylindrical model parameters and the error equation, the estimated values ​​of the cylindrical model parameters that meet the set accuracy requirements are obtained iteratively using the least squares principle. The initial axis of the air film aperture is determined from the estimated values ​​of the cylindrical model parameters.

7. The apparatus according to claim 5, characterized in that, The coordinate transformation part is configured as follows: The hole wall point cloud data after initial screening is transformed according to the rotation matrix so that the coordinates of the transformed hole wall point cloud data are perpendicular to the plane of the top view of the workpiece surface. Using the initial axis as the center, the circumference angle is divided into n equal parts. Among the transformed hole wall point cloud data corresponding to each equal angle, the hole wall point cloud data with the largest Z-axis coordinate value perpendicular to the plane of the top view of the workpiece surface is selected as the air film hole orifice point cloud data.

8. A detection system, characterized in that, The detection system includes: a numerical control device, a motion platform, a point cloud data scanning device, and a three-dimensional point cloud removal device as described in any one of claims 5 to 7; wherein... The motion platform is connected to the point cloud data scanning device and the CNC device, and controls the point cloud data scanning device to linearly acquire three-dimensional laser point cloud data of the workpiece surface. The three-dimensional point cloud removal device is connected to the point cloud data scanning device.

Images