A point set matching method, apparatus, and detection equipment based on second-order cone programming.

Abstract

This application relates to the field of industrial measurement and testing technology, specifically to a point set matching method, apparatus, and testing equipment based on second-order cone programming. This method can, to some extent, solve the problem that the pose obtained by point set matching using the least squares method leads to an overestimation of the line profile measurement result. The method includes: inputting a point cloud to be measured and a standard point cloud, and outputting the covariance matrix of the point cloud to be measured relative to the standard point cloud; constructing a rotation matrix and a translation vector; performing a rotation transformation on the point cloud to be measured, approximating the rotation matrix as linear, obtaining the maximum point set distance between the point cloud to be measured and the standard point cloud, solving for the maximum point set distance using second-order cone optimization, and obtaining and outputting the optimal rotation matrix and translation vector.

Description

Technical Field

[0001] This application relates to the field of industrial measurement and testing technology, and more specifically, to a point set matching method, apparatus, and testing equipment based on second-order cone programming. Background Technology

[0002] In industrial measurement and defect detection, collecting points on the surface of an object or workpiece to form a three-dimensional point cloud for non-contact measurement of the object's shape is a common method. Currently, the measurement of workpiece shape tolerances in industrial inspection mainly uses profile evaluation, including line profile and surface profile. Line profile primarily evaluates the variation of the actual curve relative to the ideal curve and can be used to control the shape error of planes, cylinders, cones, curved surfaces, and irregular curved surfaces. It can be applied to the entire part, multiple elements, a single surface, or individual profiles obtained from various cross-sections of a part. Point cloud profile measurement typically involves collecting point clouds of a standard workpiece and the workpiece to be measured, matching the point cloud of the workpiece to be measured with the point cloud of the standard workpiece, and calculating twice the maximum distance between the point cloud of the workpiece to be measured and the standard point cloud as the profile value.

[0003] In some contour measurement processes, the matching method used for the point set obtained by measurement is the least squares method. By considering the distance deviation of the overall position of the point cloud during the matching process, the distance deviation is minimized to obtain the rigid body transformation relationship between the point cloud of the workpiece to be measured and the point cloud of the standard workpiece.

[0004] However, since the line profile measurement value is defined as twice the maximum distance in the point set, the pose obtained by point set matching using the least squares method will lead to an overestimation of the line profile measurement result. Summary of the Invention

[0005] To address the issue that pose obtained by point set matching using the least squares method leads to overly large line profile measurement results, this application provides a point set matching method, apparatus, and detection device based on second-order cone programming.

[0006] The embodiments of this application are implemented as follows:

[0007] A first aspect of this application provides a point set matching method based on second-order cone programming, the method comprising:

[0008] Input the point cloud to be measured and the standard point cloud, decentralize and scale the point cloud to be measured and the standard point cloud respectively, and output the covariance matrix of the point cloud to be measured relative to the standard point cloud.

[0009] Construct rotation matrices and translation vectors, both of which are obtained based on the decomposition results of the covariance matrix;

[0010] A rotation transformation is performed on the point cloud to be measured, and the rotation matrix is ​​approximated as linear. The maximum distance between the point cloud to be measured and the standard point cloud is obtained. The maximum distance between the point cloud is solved by second-order cone optimization, and the optimal rotation matrix and translation vector are obtained and output.

[0011] In some embodiments, in the step of solving the maximum distance of the point set by second-order cone optimization, when the result diverges and then converges during the optimization process, the entire optimization process is recorded, and the result of the entire optimization process is reverse-searched to obtain and output the local optimal solution before divergence.

[0012] In some embodiments, the method includes the steps of decentralizing and scaling the point cloud to be measured and the standard point cloud respectively:

[0013] Calculate the center point of the point cloud to be measured, and subtract the center point from the point cloud to obtain the decentered point cloud of the point cloud to be measured;

[0014] Calculate the center point of the standard point cloud, subtract the center point from the standard point cloud to obtain the decentered point cloud of the standard point cloud;

[0015] Obtain the maximum distance between the decentered point cloud of the point cloud to be tested and the origin, and divide the decentered point cloud of the point cloud to be tested by the maximum distance to obtain the point cloud to be tested;

[0016] Obtain the maximum distance between the decentered point cloud of the standard point cloud and the origin, and divide the decentered point cloud of the standard point cloud by the obtained maximum distance to obtain the standard point cloud.

[0017] The point clouds to be tested and the standard point clouds obtained are both decentralized and uniformly sized point clouds.

[0018] In some embodiments, in the step of decomposing the obtained covariance matrix and constructing a rotation matrix and a translation vector based on the decomposition results, the method further includes:

[0019] A least-squares optimization equation is constructed for the covariance matrix, and the rotation matrix and translation vector are obtained by solving the equation.

[0020] In some embodiments, in the step of decomposing the obtained covariance matrix and constructing a rotation matrix and a translation vector based on the decomposition results, the method further includes:

[0021] Perform singular value decomposition on the covariance matrix, and construct rotation matrix and translation vector based on the singular value decomposition results.

[0022] In some embodiments, the following formula is used in the step of solving for the maximum distance of the point set using a second-order cone optimization method to obtain and output the optimal rotation matrix and translation vector:

[0023]

[0024] in, D represents the maximum distance between the point cloud to be measured and the standard point cloud; i Represents the point cloud of a standard workpiece; R represents the rotation matrix to be determined; S i t represents the point cloud of the workpiece to be measured; t represents the translation vector to be determined.

[0025] In some embodiments, the method further includes the following steps in the rotation transformation step of the point cloud to be measured:

[0026] An angle constraint factor is added. After the point cloud to be measured is rotated, the magnitude of the rotation angle around the coordinate system axis is set to be less than the angle constraint factor.

[0027] In some embodiments, the relationship between the magnitude of the rotation angle about the coordinate system axis and the angle constraint factor can be expressed by the following formula:

[0028]

[0029] Where h1, h2, and h3 are the rotation angles of the cloud to be measured around the three axes of the coordinate system, and γ is the set angle constraint factor.

[0030] A second aspect of this application provides a point set matching device based on second-order cone programming, comprising:

[0031] The preprocessing unit receives the input point cloud to be measured and the standard point cloud, performs decentralization and scale unification on the point cloud to be measured and the standard point cloud respectively, and outputs the covariance matrix of the point cloud to be measured relative to the standard point cloud.

[0032] The decomposition unit is used to construct rotation matrices and translation vectors, both of which are obtained based on the decomposition results of the covariance matrix.

[0033] The optimization unit performs a rotation transformation on the point cloud to be measured, approximates the linearization of the rotation matrix, obtains the maximum distance between the point cloud to be measured and the standard point cloud, solves the maximum distance of the point set using a second-order cone optimization method, and obtains and outputs the optimal rotation matrix and translation vector.

[0034] A third aspect of this application provides a detection device, the detection device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the point set matching method based on second-order cone programming in the above-described technical solution.

[0035] The beneficial effects of this application are as follows: By optimizing the maximum distance between the point sets of the test point cloud and the standard point cloud using a second-order cone optimization method, and outputting the rotation matrix and translation vector as the optimal transformation relationship, the accuracy of the line profile measurement value obtained according to the output transformation relationship is greater than that obtained by the line profile value obtained by the least squares method. This can reduce the over-inspection of products due to measurement errors. Furthermore, by decentralizing and unifying the scale of the test point cloud and the standard point cloud, the probability of numerical solution anomalies caused by excessively large point cloud coordinates of the test point cloud and the standard point cloud is reduced. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating a point set matching method based on second-order cone programming according to an embodiment of this application;

[0038] Figure 2 A flowchart illustrating a point set matching method based on second-order cone programming according to another embodiment of this application;

[0039] Figure 3 This is a schematic diagram of the reverse search method of the point set matching method based on second-order cone programming, according to another embodiment of this application.

[0040] Figure 4 This is a flowchart illustrating the rotation matrix optimization process of a point set matching method based on second-order cone programming, according to another embodiment of this application.

[0041] Figure 5 This is a flowchart illustrating the rotation matrix optimization process of a point set matching method based on second-order cone programming, according to another embodiment of this application.

[0042] Figure 6 This is a schematic diagram of the structure of a point set matching device based on second-order cone programming according to another embodiment of this application;

[0043] Figure 7 This is a schematic diagram showing the matching result output of a point set matching method based on second-order cone programming according to another embodiment of this application. Detailed Implementation

[0044] To make the objectives, implementation methods and advantages of this application clearer, the exemplary implementation methods of this application will be clearly and completely described below with reference to the accompanying drawings of the exemplary embodiments of this application. Obviously, the described exemplary embodiments are only some embodiments of this application, and not all embodiments.

[0045] It should be noted that the brief descriptions of terms in this application are only for the convenience of understanding the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise stated, these terms should be understood in their ordinary and common meaning.

[0046] The terms "first," "second," "third," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar or related objects or entities, and do not necessarily imply a specific order or sequence, unless otherwise specified. It should be understood that such terms are interchangeable where appropriate.

[0047] The terms “include” and “have”, and any variations thereof, are intended to cover but not exclusively include, for example, a product or device that includes a range of components is not necessarily limited to all of the components that are clearly listed, but may include other components that are not clearly listed or that are inherent to such product or device.

[0048] like Figure 1-3 As shown. Figure 1 This is a flowchart illustrating a point set matching method based on second-order cone programming according to an embodiment of this application; Figure 2 A flowchart illustrating a point set matching method based on second-order cone programming according to another embodiment of this application; Figure 3 This is a schematic diagram of the reverse search method of the point set matching method based on second-order cone programming, according to another embodiment of this application.

[0049] In some embodiments, the point set matching method based on second-order cone programming provided in this application includes:

[0050] Input the point cloud to be measured and the standard point cloud, decentralize and scale the point cloud to be measured and the standard point cloud respectively, and output the covariance matrix of the point cloud to be measured relative to the standard point cloud; wherein, the covariance matrix is ​​constructed by finding the centroid of the point cloud to be measured and the standard point cloud and the corresponding coordinates.

[0051] Construct rotation matrices and translation vectors, where both rotation matrices and translation vectors are obtained based on the decomposition results of the covariance matrix;

[0052] A rotation transformation is performed on the point cloud to be measured, and the rotation matrix is ​​approximated as linear. The maximum distance between the point cloud to be measured and the standard point cloud is obtained. The maximum distance between the point cloud is solved by second-order cone optimization, and the optimal rotation matrix and translation vector are obtained and output.

[0053] By optimizing the maximum distance between the point sets of the test point cloud and the standard point cloud using a second-order cone optimization method, and outputting the rotation matrix and translation vector as the optimal transformation relationship, the accuracy of the line profile measurement value obtained based on the output transformation relationship is greater than that obtained by the least squares method. This reduces the over-inspection of products due to measurement errors. Furthermore, by decentralizing and scaling the test point cloud and the standard point cloud, the probability of numerical solution anomalies caused by excessively large point cloud coordinates is reduced. Moreover, by performing a reverse search on the optimization process, the probability of inaccurate final optimization results can be reduced.

[0054] In some embodiments, in the step of solving the maximum distance of the point set by second-order cone optimization, when the result diverges and then converges during the optimization process, the entire optimization process is recorded, and the result of the entire optimization process is reverse-searched to obtain and output the local optimal solution before divergence.

[0055] Since the output results during the optimization process may fall into two categories, the first is that the maximum point distance gradually decreases in the later part and reaches a convergent state, with a small number of iterations, not exceeding 20; the second is that the maximum point distance diverges in the later part and the number of iterations exceeds 20. For the case where the number of iterations exceeds 20 and the maximum point set in the later part is gradually increasing, the optimal search is performed in reverse, traversing backward from the last result, and finally outputting to the minimum point. The rotation matrix and translation vector corresponding to the maximum point distance are then output as the optimal transformation relationship.

[0056] In some embodiments, to reduce the error between the point cloud to be measured and the standard point cloud, it is necessary to decentralize both the point cloud to be measured and the standard point cloud. Decentralization methods include:

[0057] Calculate the center point of the point cloud to be measured, and subtract the center point from the point cloud to obtain the decentered point cloud of the point cloud to be measured;

[0058] Calculate the center point of the standard point cloud, subtract the center point from the standard point cloud, and obtain the decentered point cloud of the standard point cloud.

[0059] Decentralized methods can be represented by the following process:

[0060] Calculate the point cloud S to be measured using the following expressions. i Center s Compared with standard point cloud Di Center D ;

[0061]

[0062]

[0063] Among them, (S) xi S yi S zi (D) represents the coordinates of the point cloud to be measured. xi D yi D zi () represents the coordinates of the standard point cloud.

[0064] Subtracting the center point from the test point cloud and the standard point cloud yields a new decentralized point cloud, expressed as follows:

[0065] S i ′=S i -Center s

[0066] D i ′=D i -Center D

[0067] Among them, S i ′ represents the centered point cloud of the point cloud to be tested, D i ′ represents the decentered point cloud of the standard point cloud.

[0068] By decentralizing the test point cloud and the standard point cloud respectively, the center point of the two point clouds can be moved to the origin, eliminating the problem of slow convergence speed caused by the large translation between the two point clouds, thereby improving the efficiency of the final output of the optimal transformation relationship.

[0069] In some embodiments, to facilitate comparison of feature values ​​between the point cloud to be measured and the standard point cloud, the methods for unifying the size of the point cloud to be measured and the standard point cloud include:

[0070] Obtain the decentralized point cloud of the point cloud to be tested and the decentralized point cloud of the standard point cloud, respectively;

[0071] Obtain the maximum distance between the decentered point cloud of the point cloud to be tested and the origin, and divide the decentered point cloud of the point cloud to be tested by the maximum distance to obtain the point cloud to be tested;

[0072] Obtain the maximum distance between the decentered point cloud of the standard point cloud and the origin, and divide the decentered point cloud of the standard point cloud by the maximum distance between the decentered point cloud of the standard point cloud and the origin to obtain the standard point cloud.

[0073] The point clouds to be tested and the standard point clouds obtained are both decentralized and uniformly sized point clouds.

[0074] This can be expressed as the following process:

[0075] Calculate the point cloud S to be measured respectively i ′ and standard point cloud D i The maximum distance M from the origin s M D ;

[0076] The point cloud to be tested S i Divide the distance M of the point cloud Si′ to be measured by the maximum distance M from the origin. s ; standard point cloud D i Divide by the standard point cloud D' i The maximum distance M from the origin D The expression is as follows:

[0077] S′ ri =S i / M s

[0078] D′ ri =D i / M s

[0079] Among them, D′ ri and S′ ri This refers to the point cloud after scaling.

[0080] The point cloud D′ obtained at this time ri Compared with standard point cloud S′ ri The maximum distance between all points in the cloud is 1, and the distances between all other points and the origin are less than 1, indicating that the point cloud D′ to be measured is... ri Compared with standard point cloud S′ ri All are located inside a sphere with a radius of 1.

[0081] By unifying the scale of the decentralized test point cloud and the standard point cloud, the stability of the numerical values ​​when calculating the data of the test point cloud and the standard point cloud can be improved, and the defect of slow convergence speed caused by the difference in size between the two point clouds can be eliminated, thereby achieving the goal of improving the efficiency of outputting the optimal transformation relationship.

[0082] In some embodiments, the step of decomposing the obtained covariance matrix and constructing the rotation matrix and translation vector based on the decomposition results further includes:

[0083] A least-squares optimization equation is constructed for the covariance matrix, and the rotation matrix and translation vector are obtained by solving the equation.

[0084] The rotation matrix and translation vector are constructed by solving the covariance matrix. The rotation matrix can also be further modified based on the covariance matrix solution, thus making it easier to obtain a more accurate translation vector.

[0085] In some embodiments, the step of decomposing the obtained covariance matrix and constructing the rotation matrix and translation vector based on the decomposition results further includes:

[0086] Perform singular value decomposition on the covariance matrix, and construct rotation matrix and translation vector based on the singular value decomposition results.

[0087] By performing singular value decomposition on the covariance matrix, the orthogonality of the column vectors in the obtained rotation matrix can be further improved, reducing the possibility that the rotation matrix obtained by using the least squares optimization equation is not orthogonal. This further improves the accuracy of constructing the rotation matrix and translation vector based on the obtained structure, and can achieve the goal of outputting the optimal transformation relationship more accurately.

[0088] In some embodiments, the following formula is used in the step of solving for the maximum distance of the point set using a second-order cone optimization method to obtain and output the optimal rotation matrix and translation vector constituting the optimal transformation relationship:

[0089]

[0090] in, Di represents the maximum distance between the point cloud to be measured and the standard point cloud; R represents the standard workpiece point cloud; S represents the rotation matrix to be determined; i t represents the point cloud of the workpiece to be measured; t represents the translation vector to be determined.

[0091] It can also be directly expressed as

[0092] The objective function for maximizing the distance between the point cloud to be measured and the standard point cloud is:

[0093] f = min(max||D) i -(RS i +t)||),

[0094] By using second-order cone programming, the rigid transformation relationship between the test point cloud and the standard point cloud is obtained. The entire optimization process is divided into multiple sub-optimization processes. The final transformation relationship is obtained by analyzing each sub-optimization process, and the optimal transformation relationship of the entire test point cloud is also obtained. At the same time, it is assumed that the test point cloud is rotated and translated in each sub-optimization process, thereby reducing the probability of local optima affecting the overall transformation relationship and ensuring that the final output transformation relationship is more consistent with the measurement results of line profile accuracy in industrial testing.

[0095] Reference Figure 4 , Figure 4 This is a flowchart of the rotation matrix optimization process for a point set matching method based on second-order cone programming, which is another embodiment of this application.

[0096] In some embodiments, the rotation transformation of the point cloud to be measured further includes:

[0097] Transform the point cloud under test based on the initial pose;

[0098] Assuming the point cloud to be measured has undergone a small-angle rotation, the rotation matrix is ​​approximately linearized and the maximum distance between the point cloud to be measured and the standard point cloud is obtained.

[0099] The objective function is to take the maximum distance of the point set as the objective function, and the maximum distance is written in the form of a second-order cone constraint. The optimization problem is then solved using the second-order cone optimization method.

[0100] Obtain the optimal rotation matrix and translation vector for the current problem.

[0101] Reference Figure 5 , Figure 5 This is a flowchart of the rotation matrix optimization process for a point set matching method based on second-order cone programming, which is another embodiment of this application.

[0102] In some embodiments, the method further includes the following steps in the rotation transformation step of the point cloud to be measured:

[0103] Transform the point cloud under test based on the initial pose;

[0104] Assuming the point cloud to be measured undergoes a small-angle rotation and an angle constraint factor is added, after the point cloud to be measured rotates, the magnitude of the rotation angle around the coordinate system axis is set to be less than the angle constraint factor.

[0105] Approximate linearization of the rotation matrix;

[0106] Find the maximum distance between the point cloud to be measured and the standard point cloud;

[0107] The objective function is to take the maximum distance of the point set as the objective function, and the maximum distance is written in the form of a second-order cone constraint. The optimization problem is then solved using the second-order cone optimization method.

[0108] Obtain the optimal rotation matrix and translation vector for the current problem.

[0109] By setting an angle constraint factor, the loss of accuracy during the linear approximation process can be minimized, thereby improving the reliability and accuracy of the optimization results. At the same time, it avoids the problem of optimization oscillations caused by excessive rotation angles, which prevent the acquisition of the optimal solution.

[0110] In some embodiments, the rotation matrix can be represented by the following expression during linear approximation.

[0111]

[0112] Where I is the identity matrix,

[0113] h1, h2, and h3 are the rotation angles of the rotation matrix R around the XYZ axes, respectively.

[0114] Let γ be the angle constraint factor, which restricts the magnitude of the rotation angle around the XYZ axes to be less than the angle constraint factor. This results in the following constraints:

[0115]

[0116] By setting an angle constraint factor, the accuracy loss of the rotation matrix during linear approximation is reduced, which facilitates the improvement of the rotation matrix and translation vector as the output of the optimal transformation relationship. This reduces the probability that the rotation matrix will oscillate during rotation, thus preventing the acquisition of the optimal solution and achieving the goal of improving the efficiency of outputting the optimal transformation relationship.

[0117] Reference Figures 6-7 , Figure 6 This is a schematic diagram of the structure of a point set matching device based on second-order cone programming according to another embodiment of this application; Figure 7 This is a schematic diagram showing the matching result output of a point set matching method based on second-order cone programming according to another embodiment of this application.

[0118] In some embodiments, based on the click matching method disclosed in the above technical solutions, this application also discloses a point set matching device based on second-order cone programming, which includes:

[0119] The preprocessing unit receives the input point cloud to be measured and the standard point cloud, performs decentralization and scale unification on the point cloud to be measured and the standard point cloud respectively, and outputs the covariance matrix of the point cloud to be measured relative to the standard point cloud.

[0120] The decomposition unit is used to construct rotation matrices and translation vectors, both of which are obtained based on the decomposition results of the covariance matrix.

[0121] The optimization unit performs a rotation transformation on the point cloud to be measured, approximates the linearization of the rotation matrix, obtains the maximum distance between the point cloud to be measured and the standard point cloud, solves the maximum distance of the point set using a second-order cone optimization method, and obtains and outputs the optimal rotation matrix and translation vector.

[0122] By optimizing the maximum distance between the point sets of the test point cloud and the standard point cloud using a second-order cone optimization method, and outputting the rotation matrix and translation vector as the optimal transformation relationship, the accuracy of the line profile measurement value obtained based on the output transformation relationship is greater than that obtained by the line profile value obtained by the least squares method. This can reduce the over-inspection of products due to measurement errors. Furthermore, by decentralizing and unifying the scale of the test point cloud and the standard point cloud, the probability of numerical solution anomalies caused by excessively large point cloud coordinates of the test point cloud and the standard point cloud is reduced.

[0123] In some embodiments, this application also discloses a detection device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to cause the at least one processor to perform a point set matching method based on second-order cone programming as described above.

[0124] The processor executes the point set matching method based on second-order cone programming in the above technical solution, including:

[0125] Input the point cloud to be measured and the standard point cloud, decentralize and scale the point cloud to be measured and the standard point cloud respectively, and output the covariance matrix of the point cloud to be measured relative to the standard point cloud; wherein, the covariance matrix is ​​constructed by finding the centroid of the point cloud to be measured and the standard point cloud and the corresponding coordinates.

[0126] Construct rotation matrices and translation vectors, where both rotation matrices and translation vectors are obtained based on the decomposition results of the covariance matrix;

[0127] A rotation transformation is performed on the point cloud to be measured, and the rotation matrix is ​​approximated as linear. The maximum distance between the point cloud to be measured and the standard point cloud is obtained. The maximum distance between the point cloud is solved by second-order cone optimization, and the optimal rotation matrix and translation vector are obtained and output.

[0128] The beneficial effects of this embodiment are that by optimizing the maximum distance between the point sets of the test point cloud and the standard point cloud using a second-order cone optimization method, and outputting the rotation matrix and translation vector as the optimal transformation relationship, the accuracy of the line profile measurement value obtained according to the output transformation relationship is greater than that obtained by the line profile value obtained by the least squares method. This can reduce the over-inspection of products due to measurement errors. Furthermore, by decentralizing and unifying the scale of the test point cloud and the standard point cloud, the probability of numerical solution anomalies caused by excessively large point cloud coordinates of the test point cloud and the standard point cloud is reduced.

[0129] For ease of explanation, the above description has been provided in conjunction with specific embodiments. However, the discussion in some embodiments above is not intended to be exhaustive or to limit the embodiments to the specific forms disclosed above. Various modifications and variations can be obtained based on the above teachings. The selection and description of the above embodiments are for the purpose of better explaining the principles and practical applications, thereby enabling those skilled in the art to better utilize the embodiments and various different variations of the embodiments suitable for specific application considerations.

Claims

1. A method for point set matching based on second order cone programming, characterized in that, The method comprises: inputting a to-be-tested point cloud and a standard point cloud, respectively decentralizing and unifying the sizes of the to-be-tested point cloud and the standard point cloud, and outputting a covariance matrix of the to-be-tested point cloud relative to the standard point cloud; constructing a rotation matrix and a translation vector, both of which are obtained based on a decomposition result of the covariance matrix; performing a rotation transformation on the to-be-tested point cloud, linearizing the rotation matrix, obtaining a point set maximum distance between the to-be-tested point cloud and the standard point cloud, solving the point set maximum distance in a second-order cone optimization manner, and obtaining and outputting an optimal rotation matrix and a translation vector; in the step of solving the point set maximum distance in the second-order cone optimization manner, when a divergence-then-convergence situation occurs in the optimization process, recording the entire optimization process and performing a reverse search on the results in the entire optimization process to obtain and output a local optimal solution before the divergence; in the step of respectively decentralizing and unifying the sizes of the to-be-tested point cloud and the standard point cloud, the method comprises: calculating a center point of the to-be-tested point cloud, subtracting the center point from the to-be-tested point cloud to obtain a decentralized point cloud of the to-be-tested point cloud; calculating a center point of the standard point cloud, subtracting the center point from the standard point cloud to obtain a decentralized point cloud of the standard point cloud; obtaining a maximum distance between the decentralized point cloud of the to-be-tested point cloud and the origin, and dividing the decentralized point cloud of the to-be-tested point cloud by the obtained maximum distance to obtain the to-be-tested point cloud; obtaining a maximum distance between the decentralized point cloud of the standard point cloud and the origin, and dividing the decentralized point cloud of the standard point cloud by the obtained maximum distance to obtain the standard point cloud; wherein the obtained to-be-tested point cloud and the standard point cloud are both decentralized and size-unified point clouds; in the step of solving the point set maximum distance in the second-order cone optimization manner and obtaining and outputting the optimal rotation matrix and the translation vector, the following formula is used: wherein, represents the maximum distance between the measured point cloud and the standard point cloud; D i represents the standard workpiece point cloud; R represents the rotation matrix to be solved; S i represents the measured workpiece point cloud; t represents the translation vector to be solved.

2. The method of claim 1, wherein the second order cone programming based point set matching method is characterized by, in the step of decomposing the obtained covariance matrix and constructing the rotation matrix and the translation vector based on the decomposition result, the method further comprises: constructing a least squares optimization equation for the covariance matrix, and obtaining the rotation matrix and the translation vector by solving the equation.

3. The point set matching method based on second-order cone programming as described in claim 1, characterized in that, in the step of decomposing the obtained covariance matrix and constructing the rotation matrix and the translation vector based on the decomposition result, the method further comprises: performing singular value decomposition on the covariance matrix, and constructing the rotation matrix and the translation vector based on the singular value decomposition result.

4. The point set matching method based on second-order cone programming as described in claim 1, characterized in that, in the step of performing a rotation transformation on the to-be-tested point cloud, the method further comprises: adding an angle restriction factor, and setting a modulus of a rotation angle around an axis of a coordinate system obtained after the to-be-tested point cloud is rotated to be less than the angle restriction factor.

5. The method of claim 4, wherein the second order cone programming based point set matching method is characterized by, The relationship between the modulus of the rotation angle around the axis of the coordinate system obtained after the to-be-tested point cloud is rotated and the angle restriction factor can be expressed by the following formula: wherein h1, h2 and h3 are respectively rotation angles of the to-be-tested point cloud around three axes of the coordinate system, and γ is the set angle restriction factor.

6. An apparatus for point set matching based on second order cone programming, characterized in that, comprise: a preprocessing unit configured to receive an input to-be-tested point cloud and a standard point cloud, respectively decentralize and unify the sizes of the to-be-tested point cloud and the standard point cloud, and output a covariance matrix of the to-be-tested point cloud relative to the standard point cloud; A decomposition unit is configured to construct a rotation matrix and a translation vector, both of which are obtained based on a decomposition result of a covariance matrix; An optimization unit is configured to perform a rotation transformation on the to-be-tested point cloud, to linearize the rotation matrix, to obtain a point set maximum distance between the to-be-tested point cloud and the standard point cloud, to solve the point set maximum distance in a second-order cone optimization manner, and to output an optimal rotation matrix and a translation vector. In the step of solving the point set maximum distance in the second-order cone optimization manner, when a divergence-then-convergence result occurs in the optimization process, the entire optimization process is recorded, and a reverse search is performed on the result in the entire optimization process to obtain and output a local optimal solution before the divergence. In the step of respectively performing decentralization and size unification on the to-be-tested point cloud and the standard point cloud, the method comprises the following steps: A center point of the to-be-tested point cloud is calculated, and the to-be-tested point cloud is subtracted by the center point to obtain a decentralized point cloud of the to-be-tested point cloud; A center point of the standard point cloud is calculated, and the standard point cloud is subtracted by the center point to obtain a decentralized point cloud of the standard point cloud; A maximum distance between the decentralized point cloud of the to-be-tested point cloud and the origin is obtained, and the decentralized point cloud of the to-be-tested point cloud is divided by the obtained maximum distance to obtain the to-be-tested point cloud; A maximum distance between the decentralized point cloud of the standard point cloud and the origin is obtained, and the decentralized point cloud of the standard point cloud is divided by the obtained maximum distance to obtain the standard point cloud; The obtained to-be-tested point cloud and the obtained standard point cloud are both point clouds after decentralization and size unification. In the step of solving the point set maximum distance in the second-order cone optimization manner and outputting the optimal rotation matrix and the translation vector, the following formula is used: wherein, represents the maximum distance between the measured point cloud and the standard point cloud; D i represents the standard workpiece point cloud; R represents the rotation matrix to be solved; S i represents the measured workpiece point cloud; t represents the translation vector to be solved.

7. A detection device, characterized by The detection device comprises at least one processor and a memory connected in communication with the at least one processor; the memory stores instructions executable by the processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the point set matching method based on the second-order cone programming according to any one of claims 1 to 5.

Images