Multi-view point cloud splicing method based on precision motion platform

Abstract

The invention relates to the technical field of three-dimensional point cloud data processing, and particularly discloses a multi-view point cloud splicing method based on a precision motion platform. A virtual plane with position marks is constructed, a relation between a camera coordinate system and a motion platform coordinate system is established by using feature point information of a calibration plate, a method for normalizing feature points of the calibration plate to a reference coordinate system is provided, and a plurality of condition constraints are established; the method provides a scheme for high-precision solving of a camera coordinate system and a platform coordinate system pose conversion model, can achieve high-precision, high-efficiency and low-calculation-cost multi-view point cloud splicing of online measurement only through one-time off-line calibration, and solves the problem that a traditional registration algorithm is difficult to carry out by calculating the characteristics of the point cloud. The splicing of the two point clouds can be realized only through repeated iteration, the number of threshold parameters is large, the efficiency is low, and local optimum is easily caused.

Description

Technical Field

[0001] This application relates to the technical field of three-dimensional point cloud data processing, and specifically discloses a multi-viewpoint cloud stitching method based on a precision motion platform. Background Art

[0002] Phase measurement profilometry is an important part of the field of optical three-dimensional measurement. Generally, it cooperates with an industrial projector through a camera to reconstruct the three-dimensional surface topography of parts through the deformation of grating stripes. It has the characteristics of low cost, high efficiency, and high precision, and is widely used in the quality inspection of industrial products such as precision stamping parts, complex molds, new energy battery packs, and printed circuit boards.

[0003] In practical applications, due to the high demand for measurement accuracy, small-field lenses are generally used in combination with a motion platform to achieve global stitching measurement of large parts. In the prior art, generally, a rough registration algorithm (such as Normal Distribution Transform, NDT) is used to calculate the initial matrix of two point clouds, and then a fine registration algorithm (such as Iterative Closest Point, ICP) and the initial matrix are used to achieve the stitching between the point clouds. However, the registration algorithm can only achieve the stitching of two point clouds through repeated iteration by calculating the features of the point clouds. There are many threshold parameters, low efficiency, and it is easy to fall into a local optimum.

[0004] In view of this, the inventor provides a multi-viewpoint cloud stitching method based on a precision motion platform to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem that the traditional registration algorithm can only achieve the stitching of two point clouds through repeated iteration by calculating the features of the point clouds. There are many threshold parameters, low efficiency, and it is easy to fall into a local optimum.

[0006] To achieve the above purpose, the basic solution of the present invention provides a multi-viewpoint cloud stitching method based on a precision motion platform, including the following steps:

[0007] Step S1: During the movement of the calibration board, multiple virtual planes with position marks are constructed with all feature points as target points;

[0008] Step S2: Arbitrarily select a calibration board coordinate system as the reference calibration board coordinate system, and normalize all feature points to the reference calibration board coordinate system;

[0009] Step S3: Combine the coordinates of the feature points on the virtual plane in the camera coordinate system and the coordinates in the motion platform coordinate system to construct an overdetermined system of equations to solve the pose transformation model;

[0010] Step S4: Establish the calculation formula for converting the relative movement amount of the precision motion platform and the initial position of the platform into the reference calibration plate coordinate system;

[0011] Step S5: Through the pose conversion model obtained in Step S3 and the calculation formula obtained in Step S4, calculate the change amount and actual three-dimensional coordinates generated by each point of the point cloud within the field of view in turn, so as to complete the stitching of the multi-field point cloud.

[0012] Furthermore, the implementation of Step S1 specifically includes the following steps:

[0013] Fasten the calibration plate to the precision motion platform, and the precision motion platform drives the calibration plate to move according to the predetermined position to form grid points on the same plane. Take all the feature points on the calibration plate as the target points, that is, multiple virtual planes with position marks are constructed from the target points.

[0014] Furthermore, in Step S2, the coordinates of the feature points in the camera coordinate system are calculated through the camera imaging model and expressed as:

[0015]

[0016] where i represents the number of measurements, i = 1, 2, 3,.......

[0017] Furthermore, in Step S2, the expression for normalizing all feature points from the camera coordinate system to the reference calibration plate coordinate system is as follows:

[0018]

[0019] In the formula, R BC and T BC respectively represent the rotation relationship and translation relationship between the reference calibration plate coordinate system and the camera coordinate system, which can be obtained by querying the camera calibration results;

[0020] The point in the precision motion platform coordinate system and the corresponding point

[0021]

[0022] In the formula, R WB and T WB respectively represent the rotation relationship and translation relationship between the platform coordinate system and the reference calibration plate coordinate system.

[0023] Furthermore, in Step S3, it is also necessary to establish the rotation matrix R WB and the translation matrix T WB between the precision motion platform coordinate system and the reference calibration plate coordinate system, and the expression is as follows:

[0024]

[0025]

[0026] Furthermore, in step S3, the constructed overdetermined system of equations is in the form of Ax = b, and the expression is as follows:

[0027]

[0028] Where 1 ≤ i ≤ M, and M ≥ 4, that is, the precision motion platform drives the calibration plate to move at least 4 positions.

[0029] Furthermore, in step S4, denote the coordinates of the platform at the first measurement as At the i-th measurement, the platform has moved relative to the first measurement by The change amount and the initial amount of the spatial points in the corresponding reference calibration plate coordinate system are respectively:

[0030]

[0031] Where

[000] represents the coordinates of the initial position of the precision motion platform in the reference calibration plate coordinate system.

[0032] Furthermore, in step S5, denote the change amount generated by each point of the point cloud in the i-th field of view as:

[0033]

[0034] Then the actual three-dimensional coordinates of the point cloud are:

[0035]

[0036] Where represents the three-dimensional coordinates output by the surface topography measurement system in the i-th field of view.

[0037] The principle and effect of this solution are as follows:

[0038] 1. Compared with the prior art, the present invention constructs a virtual plane with position marks, establishes the relationship between the camera coordinate system and the motion platform coordinate system through the feature point information of the calibration plate, proposes a method of normalizing the feature points of the calibration plate to the reference coordinate system, and establishes multiple conditional constraints, providing a solution for accurately solving the pose conversion model between the camera coordinate system and the platform coordinate system.

[0039] 2. Compared with the prior art, the present invention provides a multi-field-of-view point cloud stitching method that does not rely on point cloud feature information, avoiding the local optimum problem in the point cloud stitching process.

[0040] 3. Compared with the prior art, the present invention only needs one offline calibration to achieve high-precision, high-efficiency, and low-computation-cost multi-viewpoint cloud stitching for online measurement, solving the problem that the traditional registration algorithm can only achieve the stitching of two point clouds by calculating the features of the point cloud and iterating repeatedly, with many threshold parameters, low efficiency, and being prone to falling into local optima. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0042] Figure 1 It shows a flowchart of a multi-viewpoint cloud stitching method based on a precision motion platform proposed in an embodiment of the present application;

[0043] Figure 2 It shows a schematic diagram of the synthesis of a virtual plane with position marks proposed in an embodiment of the present application. Among them, (a) is a schematic diagram of the virtual plane formed by the predetermined positions of the platform, and (b) is a schematic diagram of the camera coordinates corresponding to the first feature point and the virtual plane;

[0044] Figure 3 It shows a schematic diagram of the multi-viewpoint cloud stitching results of different components. Among them, (a) is a schematic diagram of the multi-viewpoint cloud stitching results of a complex plastic part, (b) is a schematic diagram of the multi-viewpoint cloud stitching results of a sheet metal part, (c) is a schematic diagram of the multi-viewpoint cloud stitching results of a small wrench, and (d) is a schematic diagram of the multi-viewpoint cloud stitching results of a stainless steel trapezoidal groove part. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following will, in combination with the accompanying drawings and preferred embodiments, describe in detail the specific embodiments, structures, features, and effects of the present invention as follows.

[0046] A multi-viewpoint cloud stitching method based on a precision motion platform, as shown in the embodiments Figure 1 below, includes the following steps:

[0047] Step S1: During the movement of the calibration board, construct multiple virtual planes with position marks using all feature points as target points.

[0048] Specifically, the calibration plate is firmly connected to the precision motion platform. The precision motion platform drives the calibration plate to move according to the predetermined positions, forming grid points on the same plane. Taking all the feature points on the calibration plate as target points, that is, multiple virtual planes with position markings are constructed from the target points. As Figure 2 shown, it is a schematic diagram of the virtual plane formed by the predetermined positions of the platform and a schematic diagram of the camera coordinates corresponding to the first feature point and the virtual plane.

[0049] Step S2: Arbitrarily select a calibration plate coordinate system as the reference calibration plate coordinate system, and normalize all the feature points to the reference calibration plate coordinate system.

[0050] Specifically, calculate the coordinates of all the feature points in the camera coordinate system through the camera imaging model where i represents the number of measurements, i = 1, 2, 3,....... Then arbitrarily select a calibration plate coordinate system as the reference calibration plate coordinate system, and normalize all the feature points from the camera coordinate system to the reference calibration plate coordinate system. The expression is as follows:

[0051]

[0052] In the formula, R BC and T BC respectively represent the rotation relationship and translation relationship between the reference calibration plate coordinate system and the camera coordinate system, which can be obtained by querying the camera calibration results.

[0053] At this time, the relationship between the point in the precision motion platform coordinate system and the corresponding point in the reference calibration plate coordinate system can be expressed as:

[0054]

[0055] In the formula, R WB and T WB respectively represent the rotation relationship and translation relationship between the platform coordinate system and the reference calibration plate coordinate system.

[0056] Among them, the point on the reference calibration plate can be calculated through the above expression, while the point in the precision motion platform coordinate system can be obtained through the position sensor of the precision motion platform.

[0057] Step S3: Combine the coordinates of the feature points on the virtual plane in the camera coordinate system and the coordinates in the motion platform coordinate system to construct an overdetermined equation system as the pose conversion model.

[0058] Specifically, denote the rotation matrix R WB between the precision motion platform coordinate system and the reference calibration plate coordinate system and the translation matrix T WBThe definitions are as follows:

[0059]

[0060]

[0061] Then, by combining the coordinates in the camera coordinate system corresponding to the feature points on the virtual plane with the coordinates in the coordinate system of the precision motion platform, an overdetermined system of equations in the form of Ax = b is constructed as follows:

[0062]

[0063] In the formula, 1 ≤ i ≤ M. Since there are a total of 12 unknowns, M ≥ 4, that is, the precision motion platform drives the calibration plate to move at least 4 positions. However, due to certain errors in the coordinates of the feature points and the camera imaging model, in order to accurately solve the above parameters, M is generally much greater than 4.

[0064] Step S4: Establish a calculation formula for converting the relative movement amount of the precision motion platform and the initial position of the platform to the reference calibration plate coordinate system.

[0065] Specifically, denote the coordinates of the platform at the first measurement as At the i-th measurement, the platform has moved relative to the first measurement by Then, the change amount and the initial amount of the spatial points in the corresponding reference calibration plate coordinate system are respectively:

[0066]

[0067]

[0068] In the formula,

[000] represents the coordinates of the initial position of the precision motion platform in the reference calibration plate coordinate system.

[0069] Step S5: Through the pose conversion model obtained in Step S3 and the calculation formula obtained in Step S4, calculate the change amount and the actual three-dimensional coordinates generated by each point of the point cloud within the field of view in turn, so as to complete the stitching of the multi-field-of-view point cloud.

[0070] Denote the change amount generated by each point of the point cloud within the field of view obtained at the i-th measurement as:

[0071]

[0072] Therefore, the actual three-dimensional coordinates of the point cloud are:

[0073]

[0074] In the formula, It represents the three-dimensional coordinates output by the surface topography measurement system within the i-th field of view.

[0075] In this embodiment, using the multi-field-of-view point cloud stitching method based on a precision motion platform proposed by the present invention, the conversion model between the camera coordinate system and the platform coordinate system solved is R WB T WB = [-0.995 -0.012 0 0.606; 0.001 0.993 0 15.705; 0.133 -0.107 0 -4.931].

[0076] As Figure 3 shown, it respectively shows the schematic diagrams of the multi-field-of-view point cloud stitching results of complex plastic parts, sheet metal parts, small wrenches, and stainless steel trapezoidal groove parts.

[0077] The above is only a preferred embodiment of the present invention, and it does not impose any form of limitation on the present invention. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments by using the disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A multi-view point cloud stitching method based on a precision motion platform, characterized in that: The steps include: Step S1: During the movement of the calibration plate, multiple virtual planes with position marks are constructed with all feature points as target points; Step S2: randomly select a calibration plate coordinate system as the reference calibration plate coordinate system, and normalize all feature points to the reference calibration plate coordinate system; Step S3: combining the coordinates of the feature points on the virtual plane in the camera coordinate system and the coordinates of the motion platform coordinate system to construct an overdetermined equation group to solve the posture conversion model; Step S4: Establishing a calculation formula for converting the relative motion amount of the precision motion platform and the initial position of the platform to the reference calibration plate coordinate system; Step S5: Using the posture transformation model obtained in step S3 and the calculation formula obtained in step S4, the change amount and actual three-dimensional coordinates of each point in the point cloud within the corresponding field of view are calculated in turn, so as to complete the splicing of multi-field point clouds.

2. According to the multi-view point cloud stitching method based on a precision motion platform according to claim 1, it is characterized in that: The implementation of step S1 specifically includes the following steps: The calibration plate is tightly connected to the precision motion platform, and the precision motion platform drives the calibration plate to move according to the predetermined position to form grid points on the same plane. All feature points on the calibration plate are used as target points, that is, multiple virtual planes with position marks are constructed from the target points.

3. According to the multi-view point cloud stitching method based on a precision motion platform of claim 1, it is characterized in that: In step S2, the coordinates of the feature points in the camera coordinate system are calculated by the camera imaging model and expressed as: Wherein, i represents the number of measurements, i=1,2,3,........

4. The multi-view point cloud stitching method based on a precision motion platform according to claim 3 is characterized in that: In step S2, the expression for normalizing all feature points from the camera coordinate system to the reference calibration plate coordinate system is as follows: Where R BC and T BC They respectively represent the rotation relationship and translation relationship between the reference calibration plate coordinate system and the camera coordinate system, which can be obtained by querying the camera calibration result; Points in the coordinate system of the precision motion platform Corresponding point in the coordinate system of the reference calibration plate The relationship can be expressed as: Where R WB and T WB They respectively represent the rotation relationship and translation relationship between the platform coordinate system and the reference calibration plate coordinate system.

5. The multi-view point cloud stitching method based on a precision motion platform according to claim 1 is characterized in that: In step S3, it is also necessary to establish the rotation matrix R between the precision motion platform coordinate system and the reference calibration plate coordinate system WB With the translation matrix T WB , the expression is as follows:

6. The multi-view point cloud stitching method based on a precision motion platform according to claim 5, characterized in that: In step S3, the constructed overdetermined equation system is in the form of Ax=b, and the expression is as follows: Where, 1≤i≤M, and M≥4, that is, the precision motion platform drives the calibration plate to move at least 4 positions.

7. The multi-view point cloud stitching method based on a precision motion platform according to claim 6, characterized in that: In step S4, the coordinates of the platform during the first measurement are recorded as During the i-th measurement, the platform moves relative to the first measurement. The corresponding changes and initial values ​​of the spatial points in the reference calibration plate coordinate system are: Where [000] represents the coordinates of the initial position of the precision motion platform in the reference calibration plate coordinate system.

8. The multi-view point cloud stitching method based on a precision motion platform according to claim 7, characterized in that: In step S5, the change amount of each point in the point cloud in the i-th field of view is recorded for: The actual 3D coordinates of the point cloud are for: In the formula, Represents the three-dimensional coordinates output by the surface topography measurement system in the i-th field of view.

Images