A plate thickness detection method and device

Abstract

The present invention discloses a plate thickness detection method and device, which is suitable for measuring the global thickness of thin-walled plates. The main steps of the method include: based on the spherical fitting algorithm, using a standard sphere or a standard gauge block to normalize and calibrate the coordinate systems of two systems, laying the foundation for the rapid registration of the double-sided point cloud of the plate; collecting the double-sided point cloud of the plate through two sets of fringe projection measurement systems, and combining the PCA algorithm to correct the direction so that the plate surface point cloud is perpendicular to the Z-axis direction; with the help of the calibration results of the auxiliary target, the registration and alignment of the two point clouds are accurately achieved; based on the method of grid division combined with parallel calculation, the minimum distance between the centroid of the point cloud in the grid and all plane equations is used to characterize the distance between the two point clouds. The method described in the present invention has both robustness and efficiency because it adopts innovative methods in the links of dual-system coordinate system normalization, double-sided point cloud registration and alignment, and rapid calculation of point cloud thickness.

Description

Technical Field

[0001] The invention belongs to the technical field of optical measurement, and in particular relates to a plate thickness detection method and device. Background Art

[0002] Plates are a type of widely used industrial parts. For example, various metal raw materials are usually rolled into plates and then used as raw materials for subsequent product manufacturing. Ships, cars, rockets and other vehicles have a large number of shells, which are also a type of plate. The heat exchange carrier of most heat exchangers is also a plate structure. Plates are usually not too thick, but the cross-sectional distribution is large, which brings challenges to their thickness measurement. The thickness of the plate directly affects its strength and other performance, so it needs to be effectively measured and monitored. However, there is still a lack of efficient detection methods. Therefore, it is necessary to explore advanced technical means to detect the thickness of plates to help improve the reliability of plates, especially thin-walled plates.

[0003] CN109360267A discloses a method for fast 3D reconstruction of thin objects, which uses a depth camera to collect depth images and color images, and identifies marker points to obtain a transformation matrix. The invention has the advantage of simple operation, but its measurement range is small, and the minimum measurement thickness is only 2mm.

[0004] CN111833401A discloses a fast distance measurement method and system based on double-sided point cloud information. The invention performs rough registration through point cloud points, and performs fine registration through shape difference, and finally calculates the distance distribution of two point clouds in combination with the distance information of the sampling points. It has the advantage of relying only on three-dimensional point cloud information for registration and measurement, but is only suitable for thickness measurement of plates with the same shape on both surfaces, especially for plates with flat surfaces, and is not suitable for thickness measurement of complex curved surfaces, and the measurement efficiency is not high. Summary of the invention

[0005] In view of the shortcomings of existing global plate thickness measurement methods, especially plate thickness measurement methods with complex curved surfaces, the present invention proposes a robust, efficient and widely applicable plate thickness rapid detection method and device.

[0006] To achieve the above object, a plate thickness detection method according to the present invention comprises the following steps:

[0007] Step 1: Place the left fringe projection measurement system S L and right fringe projection measurement system S R They are arranged on both sides of the measured plate and project the left fringe measurement system S L and right fringe projection measurement system S R Carry out coordinate system normalization calibration;

[0008] Step 2: Left Fringe Projection Measurement System SL and right fringe projection measurement system S R Collect the direction-corrected surface point cloud model P of the tested panel from both sides of the panel L c And point cloud model P R c ;

[0009] Step 3: Combined with the coordinate system normalization calibration results of step 1, the surface point cloud model P of the tested panel obtained in step 2 is L c and the surface point cloud model P of the tested panel R c Perform precise registration to obtain the point cloud model P L And point cloud model P R ;

[0010] Step 4: The point cloud model P obtained in step 3 L And point cloud model P R Perform single-layer orthogonal grid division to obtain multiple spatial grids, fill all point cloud data into the spatial grids, and number the spatial grids;

[0011] Step 5: Perform plane fitting on the point cloud in the spatial grid in parallel and calculate the point cloud model P in the same spatial grid L And point cloud model P R The distance between the two plates can be used to obtain the thickness of different areas of the tested plate.

[0012] Furthermore, in step 1, a standard sphere or a standard gauge block is used for auxiliary calibration, and a homogeneous rotation and translation matrix [RT]1 is calculated by the sphere center coordinates or the plane normal vector.

[0013] Furthermore, in step 1, a spherical point cloud model or a plane point cloud model for calibration is extracted by a spherical fitting algorithm, and its spherical center coordinates or plane normal vector are calculated.

[0014] Furthermore, in step 2, principal component analysis is performed on the surface point cloud model of the panel under test collected by the left fringe projection measurement system or the right fringe projection measurement system, and the direction correction rotation and translation scheme calculated is:

[0015]

[0016] Among them, P L c Point cloud P measured by the left fringe projection measurement system L o The result after direction correction; P R c Point cloud P measured by the right fringe projection measurement system R oThe result after direction correction; [RT]2 is the eigenvector corresponding to the minimum eigenvalue of the principal component analysis result rotated to the Z-axis direction vector D z =(0,0,1) homogeneous rotation and translation matrix.

[0017] Furthermore, in step 3, based on the normalized calibration result of step 1, the point cloud model P L c And point cloud model P R c Perform precise registration, where the point cloud model P L c And point cloud model P R c The rotation and translation scheme for registration is:

[0018]

[0019] Among them, [RT]1 is the homogeneous rotation and translation matrix, P L is the point cloud model measured by the left fringe projection measurement system after registration; P R This is the point cloud model measured by the right fringe projection measurement system after registration.

[0020] Furthermore, in step 4, after dividing into a single-layer orthogonal grid, each spatial grid contains a point cloud p l (i,j) and p r (i,j), calculate each point p l =(x l ,y l ,z l ), p r =(x r ,y r ,z r ) falls into the spatial grid index number (i, j) formula is:

[0021]

[0022] Among them, p l Represents the point cloud model P measured by the left fringe projection measurement system after registration L Point in r Represents the point cloud model P measured by the right fringe projection measurement system after registration R The point in the grid; i is the index number on the X axis; j is the index number on the Y axis; w is the grid width; p l (i,j) is P L Point cloud divided into grid (i, j); p r (i,j) is P RThe point cloud is divided into the grid (i, j); (x0, y0) is the coordinate origin.

[0023] Furthermore, in step 5, the point cloud p is extracted and fitted by a plane fitting algorithm. l (i,j) and its grid neighborhood p l (i-1,j-1),p l (i-1,j),p l (i-1,j+1),p l (i,j-1),p l (i-1,j+1),p l (i+1,j-1),p l (i+1,j),p l The plane equation A of (i+1,j+1) l x+B l y+C l z+D l =0, the calculation formula of the distance d(m,n) between two point clouds is:

[0024]

[0025] Among them, m=i-1,i,i+1; n=j-1,j,j+1; (x c ,y c ,z c ) indicates p r The centroid of (i,j);

[0026] Or fitting point cloud p r (i,j) and its grid neighborhood plane equation, find the point cloud p l The distance between two point clouds is represented by the minimum distance between the centroid of (i, j) and all plane equations.

[0027] Furthermore, in step 5, each grid is traversed in parallel with the help of OMP or GPU to calculate the distance between point clouds.

[0028] A plate thickness detection device, comprising:

[0029] Normalization module, used for left fringe projection measurement system S L and right fringe projection measurement system S R Carry out coordinate system normalization calibration;

[0030] The acquisition module is used to acquire the surface point cloud model P of the tested plate after direction correction from both sides of the tested plate. L c and the surface point cloud model P of the tested panel R c ;

[0031] The precision registration module is used to combine the coordinate system normalization calibration results to calibrate the surface point cloud model P of the tested panel. L c and the surface point cloud model P of the tested panel R c Perform precise registration to obtain the point cloud model P L And point cloud model P R ;

[0032] Meshing module, used for point cloud model P L And point cloud model P R Perform single-layer orthogonal grid division to obtain multiple spatial grids, fill all point cloud data into the spatial grids, and number the spatial grids;

[0033] The thickness calculation module is used to perform plane fitting on the point cloud in the spatial grid in parallel and calculate the point cloud model P in the same spatial grid. L And point cloud model P R The distance between the two plates can be used to obtain the thickness of different areas of the tested plate.

[0034] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0035] The method disclosed in the present invention provides a new means for global thickness measurement of thin-walled plates. First, the coordinate system of the two systems is normalized and calibrated so that the double-sided point clouds of the plate can be quickly aligned. The double-sided point clouds of the plate are collected by two fringe projection measurement systems, and the direction is corrected in combination with the principal component analysis (PCA) algorithm so that the plate surface point cloud is perpendicular to the Z-axis direction, which provides a basis for the orthogonal grid division stage. With the help of the normalized calibration results of the auxiliary target, the registration and alignment of the two point clouds can be accurately achieved, which effectively simplifies the registration process. Based on the method of combining grid division with parallel calculation, the minimum distance between the centroid of the point cloud in the grid and all plane equations is used to characterize the distance between the two point clouds, which effectively improves the computational efficiency and adaptability of the thickness extraction algorithm. Since innovative methods are used in the double-sided point cloud registration and alignment, point cloud thickness rapid calculation and other links, it has both robustness and efficiency.

[0036] In actual measurement, after the coordinate systems of the two systems are normalized, each grid is traversed in parallel and combined with the orthogonal grid division technology, the global step-by-step solution is converted into a local parallel solution, which can significantly improve the measurement speed of the plate thickness.

[0037] Furthermore, the present invention performs auxiliary calibration through a standard sphere or a standard gauge block to normalize the coordinate system of the dual system, and this step only needs to be performed once in actual use, and subsequent measurements can use this calibration result to perform double-sided point cloud precise alignment. Compared with the existing fast ranging method based on double-sided point cloud information, each measurement requires two alignments from coarse alignment to fine alignment through the characteristics of the point cloud itself. The present invention greatly simplifies the double-sided point cloud alignment workflow, and has the advantages of simple operation and faster calculation speed.

[0038] Furthermore, the present invention has good environmental adaptability and a wide range of measurement objects: the existing fast ranging method based on double-sided point cloud information is only suitable for the thickness measurement of plates with flat surfaces, and is not suitable for the thickness measurement of complex curved surfaces. The present invention calculates the thickness of the grid and its neighborhood, which can effectively measure complex curved panels.

[0039] Furthermore, the present invention has a wide measurement range of thickness and high measurement accuracy: relying on the high-precision characteristics of binocular fringe projection, compared with the existing thin object rapid three-dimensional reconstruction method using a mobile laser scanner to obtain a measurement range of 2mm-30mm, the fringe projection technology has significantly improved the accuracy of the plate surface point cloud measurement, and the measurement range of the present invention can be expanded to 0.5mm-500mm. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a schematic diagram of the thin-walled plate thickness measurement system layout;

[0041] Figure 2 It is a standard sphere point cloud image;

[0042] Figure 3 It is the point cloud image of the unregistered panel;

[0043] Figure 4 This is the side view of the panel point cloud after precise registration;

[0044] Figure 5 This is a schematic diagram of grid division;

[0045] Figure 6 This is a schematic diagram of the mesh global thickness calculation results;

[0046] Figure 7 This is a schematic diagram of the module structure of the point cloud thickness measurement device provided by the present invention.

[0047] In the accompanying drawings, 1-1 is the panel under test, 1-2 is the left fringe projection measurement system, 1-3 is the right fringe projection measurement system, 2-1 is the standard sphere point cloud measured by the left fringe projection measurement system, 2-2 is the standard sphere point cloud measured by the right fringe projection measurement system, 3-1 is the board surface point cloud measured by the left fringe projection measurement system, and 3-2 is the board surface point cloud measured by the right fringe projection measurement system. DETAILED DESCRIPTION

[0048] In order to make the purpose and technical solution of the present invention clearer and easier to understand, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more. In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0050] Example 1

[0051] A plate thickness detection method comprises the following steps:

[0052] Step 1: Place the left fringe projection measurement system S L (hereinafter referred to as the left system) and the right fringe projection measurement system S R The left fringe projection measurement system and the right fringe projection measurement system (hereinafter referred to as the right system) are arranged opposite to each other, and the left fringe projection measurement system and the right fringe projection measurement system collect 3-6 images of standard balls or standard gauge blocks, and the coordinate system of the two fringe projection systems is normalized and calibrated; preferably, more than or equal to 6 images of standard balls or standard gauge blocks are collected, and the standard balls or standard gauge blocks should be evenly arranged in space.

[0053] A standard ball or a standard gauge block is used for auxiliary normalization calibration. In this embodiment, a standard ball is used for auxiliary calibration. First, the left fringe projection measurement system S LMeasure the left images of the six standard balls and obtain the point clouds 2-1 of the six standard balls measured by the left fringe projection measurement system; use the right fringe projection measurement system S R The right images of six standard balls are measured to obtain six standard ball point clouds measured by the right fringe projection measurement system 2-2. The homogeneous rotation and translation matrix [RT]1 is calculated by the spherical center coordinates or plane normal vectors, wherein the spherical point cloud model or plane point cloud model for calibration is extracted by random sampling consistency or least squares algorithm, and its spherical center coordinates or plane normal vectors are calculated.

[0054] Figure 1 The figure shows a schematic diagram of two systems arranged opposite to each other for plate measurement, wherein the left fringe projection measurement system 1-2 and the right fringe projection measurement system 1-3 respectively integrate a set of high-precision projectors and two high-precision industrial black-and-white cameras with an accuracy of 10μm-20μm. The projector is arranged between the two industrial cameras, and the three devices are fixed by tripods.

[0055] Figure 2 Six standard sphere point clouds are used for normalization of the coordinates of the two systems. The center coordinates of each standard sphere point cloud are fitted and extracted, and the homogeneous rotation and translation matrix [RT]1 is obtained using the center coordinates of the sphere.

[0056] Step 2: Use two fringe projection measurement systems to obtain the direction-corrected surface point cloud model P of the thin-walled plate from both sides of the plate. L c (Left System S L collected) and point cloud model P R c (Right System S R collected);

[0057] Measuring system S with left fringe projection L and right fringe projection measurement system S R The point cloud models are collected separately, and the principal component analysis (PCA) is performed on the surface point cloud model of the thin-walled plate collected by the right fringe projection measurement system. The direction correction rotation and translation scheme is calculated as follows:

[0058]

[0059] Among them, P L c Point cloud P measured by the left system L o The result after direction correction; P R c Point cloud P measured by the right system R oThe result after direction correction; [RT]2 is the eigenvector corresponding to the minimum eigenvalue of the principal component analysis result rotated to the Z-axis direction vector D z =(0,0,1) homogeneous rotation and translation matrix.

[0060] Figure 3 The left and right board surface point cloud images are direction-corrected but not registered. After direction correction, the board surface point cloud 3-1 measured by the left fringe projection measurement system and the board surface point cloud 3-2 measured by the right fringe projection measurement system are precisely registered in step 3.

[0061] Step 3: Combine the coordinate system normalization calibration result of step 1 with the rotation and translation matrix [RT]1 to precisely align the two point cloud models described in step 2 to obtain the point cloud model P L And point cloud model P R ;

[0062] Based on the normalized calibration results of step 1, the point cloud model P L c And point cloud model P R c Perform precise registration, point cloud model P L c And point cloud model P R c The rotation and translation scheme is:

[0063]

[0064] Among them, P L is the point cloud model measured by the left fringe projection measurement system after registration; P R This is the point cloud model measured by the right fringe projection measurement system after registration.

[0065] Figure 4 The point cloud images of the left and right surfaces of the panel after precise registration. The surface features a complex wavy shape.

[0066] Step 4: Divide the two point cloud models obtained in step 3 into single-layer orthogonal grids, and use the point cloud model P measured by the left fringe projection measurement system after registration. L And the point cloud model P measured by the right fringe projection measurement system after registration R Fill all point cloud data into the space grid and number the space grid;

[0067] After dividing into single-layer orthogonal grids, each spatial grid contains two point clouds p l (i,j) and p r (i,j), calculate each point p l =(x l ,yl ,z l ), p r =(x r ,y r ,z r ) falls into the spatial grid numbering (i, j) scheme as follows:

[0068]

[0069] Among them, p l Represents the point cloud model P measured by the left fringe projection measurement system after registration L Point in r Represents the point cloud model P measured by the right fringe projection measurement system after registration R The point in the grid; i is the index number on the X axis; j is the index number on the Y axis; w is the grid width; p l (i,j) is P L Point cloud divided into spatial grid (i, j); p r (i,j) is P R The point cloud is divided into the grid (i, j); (x0, y0) is the coordinate origin.

[0070] Figure 5 Schematic diagram of a single-layer grid after grid division.

[0071] Step 5: Use the parallel processing method to perform plane fitting on the point cloud in the spatial grid, calculate the distance between two point clouds in the same spatial grid, and obtain the grid thickness of different areas on the measured panel.

[0072] Extract and fit the point cloud p through random sampling consensus algorithm l (i,j) and its grid area p l (i-1,j-1),p l (i-1,j),p l (i-1,j+1),p l (i,j-1),p l (i-1,j+1),p l (i+1,j-1),p l (i+1,j),p l The plane equation A of (i+1,j+1) l x+B l y+C l z+D l =0, the distance d(m,n) between two point clouds is calculated as:

[0073]

[0074] Among them, m=i-1,i,i+1; n=j-1,j,j+1; (x c ,y c ,z c ) indicates p r The centroid of (i,j); in particular, the point cloud p can also be fitted r (i, j) and the plane equation of its grid area to obtain the point cloud p l The distance between two point clouds is represented by the minimum distance between the centroid of (i, j) and all plane equations.

[0075] At the same time, in this step, each grid is traversed in parallel with the help of OMP or GPU to calculate the distance between point clouds. The distance between point clouds is the thickness of the grid area of ​​the panel. The set of thicknesses of each grid area is the global thickness of the panel under test, such as Figure 6 shown.

[0076] Example 2

[0077] The present invention provides a plate thickness detection device, such as Figure 7 As shown, it includes an acquisition module, a processing module and a display module. Among them, the acquisition module is used to acquire the point cloud data of the object to be measured and transmit the acquired point cloud data to the processing module; the acquisition module includes a left fringe projection measurement system 1-2 and a right fringe projection measurement system 1-3, which are used to acquire the surface point cloud model P of the thin-walled plate after direction correction from both sides of the plate L c And the surface point cloud model of thin-walled plate P R c ;;The processing module is used to process and calculate the point cloud data of the measured object, obtain the thickness of the measured object, and transmit it to the display module; the display module is used to display the global thickness of the measured object.

[0078] The processing module comprises:

[0079] A normalization module, used for normalizing and calibrating the coordinate system of the left fringe projection measurement system and the right fringe projection measurement system;

[0080] The precise registration module is used to combine the coordinate system normalization results to align the panel surface point cloud model P L c And the plate surface point cloud model P R c Perform precise registration to obtain the point cloud model P L And point cloud model P R ;

[0081] Meshing module, used for point cloud model P L And point cloud model P RPerform single-layer orthogonal grid division, fill all point cloud data into the spatial grid, and number the grids;

[0082] Thickness calculation module, used to perform plane fitting on the point cloud in the grid in parallel and calculate the point cloud model P in the same grid L And point cloud model P R The distance between the two plates can be used to obtain the thickness of different areas of the tested plate.

[0083] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A plate thickness detection method, characterized in that: The following steps are involved: Step 1: Place the left fringe projection measurement system and right fringe projection measurement system They are arranged on both sides of the measured plate and project the left fringe measurement system and right fringe projection measurement system Carry out coordinate system normalization calibration; Step 2: Left Fringe Projection Measurement System and right fringe projection measurement system Collect the direction-corrected surface point cloud model of the tested panel from both sides of the panel and point cloud models ; Step 3: Combine the coordinate system normalization calibration results of step 1 and the surface point cloud model of the tested panel obtained in step 2 And the surface point cloud model of the tested panel Perform precise registration to obtain a point cloud model and point cloud models ; Step 4: The point cloud model obtained in step 3 and point cloud models Perform single-layer orthogonal grid division to obtain multiple spatial grids, fill all point cloud data into the spatial grids, and number the spatial grids; Step 5: Perform plane fitting on the point cloud in the spatial grid in parallel and calculate the point cloud model in the same spatial grid and point cloud models The distance between the two plates can be used to obtain the thickness of different areas of the tested plate; In step 2, principal component analysis is performed on the surface point cloud model of the panel under test collected by the left fringe projection measurement system or the right fringe projection measurement system, and the direction correction rotation and translation scheme calculated is: （1） in, Point cloud measured by the left fringe projection measurement system The result after direction correction; Point cloud measured by the right fringe projection measurement system The result after direction correction; The eigenvector corresponding to the minimum eigenvalue of the principal component analysis result is rotated to Axis direction vector The homogeneous rotation and translation matrix of ; In step 3, based on the normalized calibration result of step 1, the point cloud model and point cloud models Perform precise registration, where the point cloud model and point cloud models The rotation and translation scheme for registration is: （2） in, is the homogeneous rotation and translation matrix, It is the point cloud model measured by the left fringe projection measurement system after registration; It is the point cloud model measured by the right fringe projection measurement system after registration; In step 4, after dividing into a single layer of orthogonal grids, each spatial grid contains a point cloud and , calculate each point , The index number of the space grid into which it falls The formula is: （3） in, Represents the point cloud model measured by the left fringe projection measurement system after registration The point in Represents the point cloud model measured by the right fringe projection measurement system after registration The point in For The index number on the axis; For The index number on the axis; is the grid width; for Divide into grid Point cloud in for Divide into grid Point cloud in is the coordinate origin; In step 5, the point cloud is extracted and fitted by a plane fitting algorithm. and its grid neighborhood , , , , , , , The plane equation , the calculation formula of the distance d(m,n) between two point clouds is: （4） in, ; ; express The centroid of or Fit Point Cloud and the plane equation of its grid neighborhood to obtain the point cloud The distance between two point clouds is characterized by the minimum distance between the center of mass and all plane equations.

2. A plate thickness detection method according to claim 1, characterized in that: In step 1, a standard ball or a standard gauge block is used for auxiliary calibration, and the homogeneous rotation and translation matrix is ​​calculated by the spherical center coordinates or the plane normal vector. .

3. A plate thickness detection method according to claim 2, characterized in that: In the step 1, a spherical point cloud model or a plane point cloud model for calibration is extracted by a spherical fitting algorithm, and its spherical center coordinates or plane normal vector are calculated.

4. A plate thickness detection method according to claim 1, characterized in that: In step 5, each grid is traversed in parallel with the help of OMP or GPU to calculate the distance between point clouds.

5. A plate thickness detection device, used to implement the method of claim 1, characterized in that: include: Normalization module for left fringe projection measurement system and right fringe projection measurement system Carry out coordinate system normalization calibration; The acquisition module is used to acquire the surface point cloud model of the tested plate after direction correction from both sides of the tested plate. And the surface point cloud model of the tested panel ; The precision registration module is used to combine the coordinate system normalization calibration results to calibrate the surface point cloud model of the tested panel. And the surface point cloud model of the tested panel Perform precise registration to obtain a point cloud model and point cloud models ; Meshing module for point cloud models and point cloud models Perform single-layer orthogonal grid division to obtain multiple spatial grids, fill all point cloud data into the spatial grids, and number the spatial grids; Thickness calculation module, used to perform plane fitting on point clouds in spatial grids in parallel and calculate point cloud models in the same spatial grid and point cloud models The distance between the two plates can be used to obtain the thickness of different areas of the tested plate.

Images