CN-122023482-A - Machining allowance calculation method based on three-dimensional point cloud registration

CN122023482ACN 122023482 ACN122023482 ACN 122023482ACN-122023482-A

Abstract

A machining allowance calculation method based on three-dimensional point cloud registration relates to the technical field of machining. The invention aims to solve the problems of low calculation accuracy and low efficiency of the existing machining allowance calculation method. The method comprises the steps of obtaining an ideal position and pose point cloud P of a part and a point cloud Q of the part to be registered, performing rough registration by using the P and Q to obtain a rough registration result, dividing points in the ideal position and pose point cloud P of the part into a boundary local point and a non-boundary local point, performing fine registration by using the rough registration result, the boundary local point and the non-boundary local point to obtain an end point cloud result, and obtaining machining allowance corresponding to each point in the end point cloud result. The invention obtains the machining allowance of the curved surface of the part relative to the ideal curved surface.

Inventors

Sui tianyi
CHEN ZIHUI
CHANG JIAQI
DONG XUAN
ZHANG TONG

Assignees

天津大学

Dates

Publication Date: 20260512
Application Date: 20260126

Claims (10)

1. A machining allowance calculation method based on three-dimensional point cloud registration is characterized by comprising the following specific processes: firstly, acquiring an ideal pose point cloud P of a part and a point cloud Q of the part to be registered, and performing coarse registration by using the P and the Q to obtain a coarse registration result; Dividing points in the ideal pose point cloud P of the part into a boundary local point and a non-boundary local point; step three, performing fine registration by using the coarse registration result obtained in the step one, the boundary local point and the non-boundary local point obtained in the step two, and obtaining the final point cloud result ; Step four, obtaining the final point cloud result Machining allowance corresponding to each point.
2. The method for calculating the machining allowance based on the three-dimensional point cloud registration of claim 1, wherein the method for calculating the machining allowance is characterized in that the ideal pose point cloud P of the part and the point cloud Q of the part to be registered are obtained in the first step, and coarse registration is carried out by utilizing the P and the Q to obtain a coarse registration result, specifically: the method comprises the steps of obtaining an ideal pose point cloud P of a part and a point cloud Q of the part to be registered, and specifically comprises the following steps: Performing discretization point cloud processing on a three-dimensional CAD model of a standard part to obtain an ideal pose point cloud P, and obtaining a point cloud Q of the part to be registered by using a three-dimensional scanner; Step two, acquiring ideal pose point cloud centroid by utilizing ideal pose point cloud P of part Acquiring point cloud centroids of parts to be registered by utilizing point cloud Q of parts to be registered And respectively utilize And Performing decentration treatment on the ideal pose point cloud of the part and the point cloud of the part to be registered to obtain the decentered ideal pose point cloud of the part and the decentered point cloud of the part to be registered; Step one, acquiring ideal characteristic point clouds by using the ideal position point clouds of the part subjected to decentralization and the point clouds of the part to be registered subjected to decentralization respectively And feature point cloud to be registered ; Step one, four, utilizing ideal characteristic point cloud And feature point cloud to be registered And obtaining a rough matching result.
3. The method for calculating machining allowance based on three-dimensional point cloud registration according to claim 2, wherein ideal pose point cloud centroid is obtained by utilizing ideal pose point cloud P of part in the first step and the second step Acquiring point cloud centroids of parts to be registered by utilizing point cloud Q of parts to be registered And respectively utilize And Performing decentration treatment on the ideal pose point cloud of the part and the point cloud of the part to be registered to obtain the ideal pose point cloud of the part after decentration and the point cloud of the part to be registered after decentration, wherein the ideal pose point cloud of the part and the point cloud of the part to be registered after decentration are specifically as follows: acquiring ideal pose point cloud centroid by utilizing ideal pose point cloud P of part Acquiring point cloud centroids of parts to be registered by utilizing point cloud Q of parts to be registered The method specifically comprises the following steps: Wherein, the Is the number of points in the ideal pose point cloud of the part, Is the index of the midpoint of the ideal pose point cloud of the part, Is the number of points in the point cloud of the part to be registered, Is the index of the point cloud midpoint of the part to be registered, Is the first position and pose point cloud of the ideal part A point of the light-emitting diode is located, Is the first part point cloud to be registered A plurality of points; Step two by two, respectively utilize And Performing decentration treatment on the ideal pose point cloud of the part and the point cloud of the part to be registered to obtain the ideal pose point cloud of the part after decentration and the point cloud of the part to be registered after decentration, wherein the ideal pose point cloud of the part and the point cloud of the part to be registered after decentration are specifically as follows: Wherein, the Is after being decentered , Is after being decentered 。
4. The method for calculating machining allowance based on three-dimensional point cloud registration according to claim 3, wherein the ideal characteristic point cloud is obtained from the decentralised ideal position point cloud of the part and the decentralised ideal position point cloud of the part to be registered in the third step And feature point cloud to be registered The method specifically comprises the following steps: Step one, three and one, obtaining the distance between the point in the ideal pose point cloud of the part after decentralization and the mass center And spatial local density The method specifically comprises the following steps: Wherein, the Is that The distance to the centroid of the image is, Is a sign of a modular length and, Is the first position and pose point cloud of the ideal part A point of the light-emitting diode is located, Is the local density of points in the ideal pose point cloud of the part after the decentralization, Is the minimum value; step one, three and two, obtaining the distance between points in the point cloud of the part to be registered and the centroid after decentralization And spatial local density The method specifically comprises the following steps: Wherein, the Is that The distance to the centroid of the image is, Is the first part point cloud to be registered A plurality of points; step one, three and three, use 、、 And Acquiring ideal characteristic point cloud And feature point cloud to be registered The method specifically comprises the following steps: first, will And (3) with And used as the ideal pose point cloud of the part after the decentralization Scoring each point in the ideal pose point cloud of the centered part according to the descending order, and sorting the front part The points corresponding to the scores form an ideal characteristic point cloud ; Then, will And (3) with And used as the first part point cloud to be registered after decentralization Scoring each point in the centered part point cloud to be registered according to the descending order of the scores of the points, and forwarding the scores of the points to be registered The points corresponding to the scores form feature point cloud to be registered 。
5. The method for computing the machining allowance based on three-dimensional point cloud registration according to claim 4, wherein ideal characteristic point cloud is utilized in the fourth step And feature point cloud to be registered The rough matching result is obtained specifically as follows: Step one, four and one, obtaining ideal characteristic point cloud Centroid of (2) And feature point cloud to be registered Centroid of (2) And utilize And Performing decentration on the feature points in the ideal feature point cloud and the feature points in the feature point cloud to be registered to obtain an ideal feature point cloud after decentration and a feature point cloud to be registered after decentration; the ideal characteristic points after the decentralization and the characteristic points to be registered after the decentralization are obtained by the following modes: Wherein, the Is the first after the decentralization The number of the ideal characteristic points is one, Is the first The number of the ideal characteristic points is one, Is an ideal characteristic point cloud Is used to determine the centroid of the (c), Is the feature point cloud to be registered Is used to determine the centroid of the (c), Is the first after the decentralization The feature points to be registered are arranged in the registration table, Is the first The feature points to be registered are arranged in the registration table, Is the total number of feature points, Is a feature point label; Step one, four and two, acquiring a decentralised ideal feature matrix by using the decentralised ideal feature point cloud and the decentralised feature point cloud to be registered And the feature matrix to be registered after the decentralization By using And Constructing covariance matrix Singular value decomposition is carried out on the covariance matrix to obtain a left singular vector matrix And right singular vector matrix ; The decentralised ideal feature matrix The middle row represents the coordinates of one point in the ideal characteristic point cloud after decentralization; The feature matrix to be registered after the decentralization The middle row represents the coordinates of one point in the feature point cloud to be registered after the decentralization; The ideal characteristic matrix after the decentralization is utilized And the feature matrix to be registered after the decentralization Constructing covariance matrix The method specifically comprises the following steps: Wherein, the Is a transpose; Singular value decomposition is carried out on the covariance matrix, and the singular value decomposition is specifically as follows: Wherein, the Is a matrix of left singular vectors and, Is a matrix of right singular vectors and, Is a singular value matrix; Step one, four and three, utilizing left singular vector matrix And right singular vector matrix Obtaining coarse registration rotation matrices And coarse registration translation matrix The method specifically comprises the following steps: Step one, four and four, utilizing coarse registration rotation matrix And coarse registration translation matrix Converting the point cloud Q of the part to be registered to obtain a rough registration result, specifically: Wherein, the Is the point cloud of the part to be registered after rough registration.
6. The method for calculating machining allowance based on three-dimensional point cloud registration of claim 5, wherein the dividing of points in the ideal pose point cloud P of the part into boundary local points and non-boundary local points in the second step is achieved by adopting an alpha-shape algorithm.
7. The method for computing the machining allowance based on the three-dimensional point cloud registration of claim 6, wherein the coarse registration result obtained in the first step and the boundary local point and the non-boundary local point obtained in the second step are used for fine registration in the third step to obtain an end point cloud result The method specifically comprises the following steps: Step three, initializing the iteration times Initializing a rotation matrix Initializing a translation matrix Initializing a registration matrix Initializing learning rate ; Wherein, the Is a constant; the said Based on preset initial unit quaternion The preparation method comprises the following steps: Wherein, the Is a preset initial unit quaternion, Is an element value in a preset unit quaternion, Is the initial rotation matrix; the said Based on The preparation method comprises the following steps: Wherein, the Is the initial registration matrix of the image, Is an initial translation matrix; Step three, searching by utilizing KD-Tree Midpoint (midpoint) At the position of Corresponding point in (3) Acquisition of Normal vector of (2) And curvature of By using 、、 And Obtaining a homogenized elevation function value ; Step III, utilize Obtaining a loss function value Will lose the function And a preset objective loss function value Comparison will And maximum number of iterations Comparing, if And is also provided with Then use the learning rate Updating the rotation matrix and the translation matrix to obtain an updated rotation matrix And an updated translation matrix Then executing the third and fourth steps, otherwise, directly obtaining the registration matrix Order-making And outputting the final cloud result ; Step three and four, utilizing the updated rotation matrix And an updated translation matrix Obtaining an updated registration matrix The method specifically comprises the following steps: Step three, five, judge Whether or not it is an integer, if Is an integer, let Then let the And returning to the third step, otherwise, letting Then let the And returning to the third step; Wherein, the 。
8. The method for calculating machining allowance based on three-dimensional point cloud registration according to claim 7, wherein the three-dimensional point cloud registration is utilized in the third step 、、 And Obtaining a homogenized elevation function value The method specifically comprises the following steps: Wherein, the Is the length of the mould, and the mould is the mould, Is the absolute value of the absolute value, And Is the weight coefficient of the weight of the object, Is a dot The radial distance to the local sea level, Is a dot To the point Is a euclidean distance of (c).
9. The method for calculating machining allowance based on three-dimensional point cloud registration according to claim 8, wherein the three-dimensional point cloud registration is utilized Obtaining a loss function value The method specifically comprises the following steps: the utilization learning rate Updating the rotation matrix and the translation matrix to obtain an updated rotation matrix And an updated translation matrix The method specifically comprises the following steps: The updated rotation matrix Based on updated quaternion Obtaining; the updated quaternion Obtained by: Wherein, the Is a quaternion The gradient is such that, Is the first The quaternion of the round of iteration, Is the first Quaternion of round iteration; the updated translation matrix Obtained by: Wherein, the Is a translational gradient that is set to a constant velocity, Is the first The translation matrix of the round of iteration, Is the first A translation matrix of round iterations.
10. The method for calculating machining allowance based on three-dimensional point cloud registration according to claim 9, wherein the final point cloud result is obtained in the fourth step The machining allowance corresponding to each point is specifically as follows: Step four, cloud results at the final point Middle manual pick-up Multiple points, each picking point is obtained by referring to the external normal vector The method specifically comprises the following steps: Wherein, the Is the first The individual pick-up points refer to the outer normal vector, , Is the label of the pick-up point, Is the total number of pick-up points, Is the first The normal vector of the individual pick-up points, Is the first Individual pick point orientations Is used for the reference vector of (a), Is the first The coordinates of the individual pick-up points, Is the final point cloud result The geometric center coordinates of the bounding box, Is that The minimum value of the abscissa among the coordinates of all points in (a), Is that The abscissa maximum value among the coordinates of all points in (a), Is that The minimum value of the ordinate among the coordinates of all points in (a), Is that The maximum value of the ordinate among the coordinates of all points in (a), Is that The minimum value of the vertical coordinates among the coordinates of all the points in (a), Is that The maximum value of vertical coordinates in the coordinates of all points in the (3); step four, acquiring by using PCA algorithm Normal vector of other points except the pick-up point and obtain If the included angle between the normal vector of the current point and the reference external normal vector closest to the current point is larger than a preset angle, reversing the normal vector of the current point, marking the reversed normal vector of the current point, and adding the reversed normal vector of the current point into a set Otherwise, directly marking the normal vector of the current point and adding the normal vector of the current point into the set ; Step IV, based on The direction of the normal vector is obtained Perpendicular distance from the point in (c) to the surface of the three-dimensional CAD model of the standard part Will be As the machining allowance of the current point.

Description

Machining allowance calculation method based on three-dimensional point cloud registration Technical Field The invention relates to the technical field of machining, in particular to a machining allowance calculation method based on three-dimensional point cloud registration. Background The polishing finish machining of the curved surface part is a core process for achieving the standard of the surface quality and geometric precision of the complex curved surface workpiece in the field of precision mechanical manufacturing, and the machining object of the polishing finish machining is widely covered with key parts such as aerospace engine blades, precision optical lenses, mold cavities, medical instrument joint assemblies and the like. In the polishing finishing process of curved surface parts, in order to ensure polishing quality and surface consistency, geometric deviation and machining allowance between the surface of a workpiece and a design model are required to be accurately evaluated before polishing. At present, real-point cloud data of the surface of a workpiece is mainly obtained through three-dimensional scanning and the like, and the real-point cloud data is registered with a CAD model or a nominal point cloud based on an Iterative Closest Point (ICP) method, so that the distance from the surface of the workpiece to a design curved surface is calculated, and the distance is used as machining allowance. However, since the curved surface polishing object is mostly a free curved surface or a complex curved surface part, the real-point cloud often has the conditions of noise, local missing, uneven density and large curvature change, and the traditional point cloud registration based on the Iterative Closest Point (ICP) method is sensitive to the initial pose and the data quality, the problems of unstable registration or accumulated local errors are easy to occur, and the accuracy of machining allowance calculation is low. In addition, the time consumption of the large-scale point cloud nearest neighbor search and iterative optimization is high in dependence on initial pose, parameter adjustment still exists in field application, and manual parameter adjustment can increase the point cloud registration time, so that the machining allowance calculation efficiency is low. Disclosure of Invention The invention provides a machining allowance calculation method based on three-dimensional point cloud registration, which aims to solve the problems of low calculation accuracy and low efficiency of the existing machining allowance calculation method. A processing allowance calculation method based on three-dimensional point cloud registration specifically comprises the following steps: firstly, acquiring an ideal pose point cloud P of a part and a point cloud Q of the part to be registered, and performing coarse registration by using the P and the Q to obtain a coarse registration result; Dividing points in the ideal pose point cloud P of the part into a boundary local point and a non-boundary local point; step three, performing fine registration by using the coarse registration result obtained in the step one, the boundary local point and the non-boundary local point obtained in the step two, and obtaining the final point cloud result ; Step four, obtaining the final point cloud resultMachining allowance corresponding to each point. Further, the ideal pose point cloud P of the part and the point cloud Q of the part to be registered are obtained in the first step, and coarse registration is performed by using the point cloud P and the point cloud Q to obtain a coarse registration result, specifically: the method comprises the steps of obtaining an ideal pose point cloud P of a part and a point cloud Q of the part to be registered, and specifically comprises the following steps: Performing discretization point cloud processing on a three-dimensional CAD model of a standard part to obtain an ideal pose point cloud P, and obtaining a point cloud Q of the part to be registered by using a three-dimensional scanner; Step two, acquiring ideal pose point cloud centroid by utilizing ideal pose point cloud P of part Acquiring point cloud centroids of parts to be registered by utilizing point cloud Q of parts to be registeredAnd respectively utilizeAndPerforming decentration treatment on the ideal pose point cloud of the part and the point cloud of the part to be registered to obtain the decentered ideal pose point cloud of the part and the decentered point cloud of the part to be registered; Step one, acquiring ideal characteristic point clouds by using the ideal position point clouds of the part subjected to decentralization and the point clouds of the part to be registered subjected to decentralization respectively And feature point cloud to be registered; Step one, four, utilizing ideal characteristic point cloudAnd feature point cloud to be registeredAnd obtaining a rough matching result. Further, in the step two,