Thin-walled sleeve measurement method, apparatus, and device based on calibration registration

Abstract

This invention relates to the fields of precision manufacturing and automated testing technology. Addressing the problems of inaccurate plane fitting and low measurement accuracy caused by poor point cloud quality on the upper surface of thin-walled components in existing technologies, this invention proposes a method, apparatus, and equipment for measuring thin-walled sleeves based on calibration and registration. The method includes: an offline calibration stage where point clouds of standard parts are acquired through 3D scanning or assisted measurement, and a high-precision calibration model is established; an online registration stage where the point cloud to be measured is acquired, and using high-quality features from non-measurement areas such as the outer wall, the point cloud to be measured is registered to the calibration model using a point cloud registration algorithm, achieving reference transfer; the point cloud of the measurement area is projected onto the calibration plane for circle fitting and parameter calculation; and the true geometric parameters in the original coordinate system are output through inverse transformation. This invention adopts a two-stage architecture of "offline calibration + online registration," replacing local fitting with overall registration and utilizing overall features to achieve reference transfer, thereby improving the measurement accuracy and reliability of key dimensions of thin-walled sleeve-type workpieces.

Description

Technical Field

[0001] This invention belongs to the field of precision manufacturing and automated testing technology, specifically relating to a method, apparatus and equipment for measuring thin-walled sleeves based on calibration and registration. Background Technology

[0002] With the rapid development of 3D machine vision and precision measurement technologies, high-precision dimensional inspection of industrial parts based on 3D point clouds, such as the acquisition of parameters like roundness, hole diameter, and coaxiality, has become a key aspect of intelligent manufacturing and quality control. Thin-walled sleeve-type parts (such as bearing sleeves, hydraulic cylinder barrels, and bushings) are particularly important inspection targets due to their widespread application in aerospace, precision instruments, and hydraulic transmission. The measurement of critical geometric dimensions for these parts heavily relies on the accurate extraction of parameters such as the center and radius of the annular region (central hole or outer circle) on their upper surface.

[0003] Traditional contact measurement methods (such as coordinate measuring machines) offer high accuracy, but suffer from limitations such as low efficiency, potential damage to workpiece surfaces, and difficulty in integration into automated production lines. Therefore, non-contact 3D vision measurement technologies, such as structured light and laser scanning, have become the mainstream solution for industrial online inspection due to their high efficiency and flexibility.

[0004] Currently, the industry generally adopts the following standard procedure for 3D visual measurement of thin-walled sleeve-type parts: First, complete point cloud data of the workpiece is acquired through a 3D scanning device; second, a subset of the point cloud of the upper surface (annular region) to be measured is segmented from the point cloud, and a reference plane is calculated using this subset through a statistical fitting algorithm (such as least squares method or RANSAC); then, the 3D point cloud of the upper surface region is vertically projected onto the fitted reference plane, transforming it into a 2D point set; finally, a circle is fitted on the 2D projection plane to extract target parameters such as the center coordinates and radius.

[0005] However, for thin-walled sleeves, a special type of part, the aforementioned standard procedure has fundamental accuracy defects. Because the upper surface to be measured on a thin-walled sleeve is typically a narrow annular area only a few millimeters wide, and often due to machining textures, reflective properties, or the presence of oil contamination, the point cloud data obtained from scanning this area is of poor quality, exhibiting high noise, sparse point clouds, or localized missing points. Performing planar fitting based on this data presents the following three interrelated serious problems:

[0006] The accuracy of initial plane fitting is severely limited. The accuracy of plane fitting algorithms directly depends on the quality and geometric distribution of the input point cloud. Point clouds provided by narrow annular regions not only have limited data volume and poor quality, but their spatial distribution is also not an ideal, complete plane. This leads to unavoidable large errors in the spatial position and normal direction of the reference plane obtained by any statistical fitting algorithm. This error stems from the inherent defects of low-quality raw data and is a fundamental bottleneck that is difficult to eliminate by optimizing the fitting algorithm itself.

[0007] Errors are propagated and amplified in subsequent steps. The inaccurate reference plane generated in the above steps directly leads to systematic deviations in the next critical step—"point cloud projection." The actual 3D point cloud is incorrectly projected onto a "virtual plane" with angular and positional errors, rather than the true theoretical plane. This projection operation directly introduces and solidifies the errors from the previous plane fitting into the subsequent 2D point set data used for circle fitting, thus propagating the error.

[0008] The final measurement results are distorted. Because the input data of the circle fitting algorithm—the two-dimensional projection point set—already includes geometric distortion in the previous steps, even with a high-precision circle fitting algorithm, the calculated center coordinates, radius, and other final parameters will still deviate significantly from the actual geometric dimensions of the workpiece. For thin-walled sleeves with tolerance requirements often at the micrometer level, this deviation is sufficient to lead to unreliable measurement results, causing misjudgments or rework, and failing to meet the stringent requirements of repeatability and accuracy for modern industrial online inspection.

[0009] In summary, the core bottleneck of existing technical solutions lies in their structural deficiency of relying on initial plane fitting of low-quality, narrow annular point clouds. This is not a problem that can be optimized by local algorithms, but rather a systemic inadequacy of the traditional "fit first, then project" measurement process when dealing with the special features of thin-walled parts. Therefore, there is an urgent need in this field for an innovative method that can fundamentally avoid or correct this initial plane fitting error in order to achieve stable, reliable, and high-precision measurement of the key geometric dimensions of thin-walled sleeve-type parts. Summary of the Invention

[0010] To address the aforementioned technical problems, this invention proposes a method, apparatus, and device for measuring thin-walled sleeves based on calibration and registration. This involves establishing or referencing a digitized calibration model containing high-precision reference features, using its coordinate system as a unified reference; aligning the point cloud to be measured to the reference space using a high-precision registration algorithm (e.g., ICP); and finally performing feature projection and fitting on the reference surface. This approach achieves precise measurement through a technical route that integrates physical measurement and CAD dual calibration.

[0011] This invention provides a method for measuring thin-walled sleeves based on calibration registration, comprising: Step S100: Perform offline calibration, establish a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establish the spatial coordinate system of the calibration model as the reference coordinate system; the calibration model includes at least a high-precision reference plane and reference circle feature defined by the upper surface of the thin-walled sleeve. Step S200: Perform online registration, including acquiring the measurement point cloud of the thin-walled sleeve to be measured, and registering the measurement point cloud to the reference coordinate system by determining the spatial transformation relationship of transforming the measurement point cloud to the reference coordinate system. Step S300: Project the points corresponding to the upper surface in the measurement point cloud registered in the reference coordinate system onto the reference plane; perform circle fitting on the projected points on the reference plane to obtain the fitting circle parameters in the reference coordinate system. Step S400: Based on the spatial transformation relationship, the center coordinates of the fitted circle parameters are inversely transformed to the original coordinate system where the measured point cloud is located, and the measurement results are output.

[0012] Preferably, in step S100, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design includes: Obtaining the calibration point cloud of the reference thin-walled sleeve includes using a 3D scanning device to collect complete point cloud data of a known qualified thin-walled sleeve as a reference, which is used as the calibration point cloud; the calibration point cloud completely covers the point cloud data of the upper surface, outer cylindrical surface, inner hole wall and lower end surface of the reference thin-walled sleeve. A calibration model containing high-precision reference plane and reference circle features is established based on the calibration point cloud.

[0013] Preferably, in step S100, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design includes: The upper surface of the reference thin-walled sleeve was measured using a contact probe to obtain measurement data; The calibration model parameters of the high-precision reference plane and reference circle features are obtained directly based on the measurement data fitting; the calibration model parameters include at least the equation parameters of the reference plane and the calibration center and calibration radius of the reference circle, and the flatness of the upper surface is obtained during the measurement process to verify the accuracy of the reference plane.

[0014] Preferably, in step S100, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establishing the spatial coordinate system of the calibration model as the reference coordinate system, includes: The CAD design model of the thin-walled sleeve to be tested is directly used as the calibration model for the theoretical design geometric features of the thin-walled sleeve; Establishing the spatial coordinate system of the calibration model as the reference coordinate system includes: By scanning a reference thin-walled sleeve of known dimensions, a fixed spatial transformation relationship between the coordinate system of the three-dimensional scanning device and the coordinate system of the CAD design model is calculated and determined, and the coordinate system of the CAD design model is established as the reference coordinate system.

[0015] Preferably, in step S200, determining the spatial transformation relationship for registering the measured point cloud to the reference coordinate system includes: Using the same 3D scanning equipment and settings as in the offline calibration phase, complete point cloud data of the thin-walled sleeve under test was acquired as the measurement point cloud. The measured point cloud and the calibration point cloud are registered in three dimensions, and the optimal spatial transformation matrix between the measured point cloud and the calibration point cloud in the reference coordinate system is calculated.

[0016] Preferably, in step S200, determining the spatial transformation relationship for registering the measured point cloud to the reference coordinate system includes: The measured point cloud is transformed to a reference coordinate system using the fixed spatial transformation relationship. By performing model-based iterative nearest-point registration between the measured point cloud transformed to the reference coordinate system and the surface of the CAD design model, the corresponding correction transformation is calculated. By combining the fixed spatial transformation relationship with the corrected transformation, the spatial transformation relationship between the measured point cloud and the reference coordinate system is obtained.

[0017] On the other hand, the present invention also provides a thin-walled sleeve measuring device based on calibration registration, comprising: The first module is used for offline calibration, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establishing the spatial coordinate system of the calibration model as the reference coordinate system; the calibration model includes at least a high-precision reference plane and reference circle feature defined by the upper surface of the thin-walled sleeve. The second module is used for online registration, including acquiring the measurement point cloud of the thin-walled sleeve to be measured, and registering the measurement point cloud to the reference coordinate system by determining the spatial transformation relationship of transforming the measurement point cloud to the reference coordinate system. The third module is used to project the points corresponding to the upper surface in the measurement point cloud registered in the reference coordinate system onto the reference plane; and to perform circle fitting on the projected points on the reference plane to obtain the fitting circle parameters in the reference coordinate system. The fourth module is used to inversely transform the center coordinates of the fitted circle parameters to the original coordinate system of the measured point cloud according to the spatial transformation relationship, and output the measurement results.

[0018] Compared with the prior art, the beneficial effects of the present invention include: (1) Significantly improved measurement accuracy: By registering to the calibration model (whether from fitting high-standard parts or directly referencing CAD models), the initial plane fitting error is completely avoided, and the high-precision known plane is directly used. As a projection reference, it eliminates the reference surface error caused by poor point cloud quality on the upper surface from the root.

[0019] (2) Improved stability and repeatability of the measurement system: Once the calibration model M is established in the offline stage, its geometric features (plane, center) as the reference are fixed and highly accurate, and can be used repeatedly. The consistency of subsequent measurements mainly depends on the accuracy of point cloud registration, and the registration algorithm utilizes the overall and rich geometric features of the part (such as inner and outer cylindrical surfaces), and its stability and repeatability are far higher than the traditional method of fitting based only on a narrow upper surface area.

[0020] (3) Balancing measurement accuracy and online efficiency: This invention places the most critical task of "establishing a high-precision benchmark" in a one-time offline calibration stage, allowing for more time and manpower to ensure optimal results. In the online measurement stage, the process is fully automated (automatic acquisition, registration, calculation, and output), with single-piece measurement efficiency comparable to or even higher than traditional methods, while achieving a qualitative leap in the accuracy and reliability of the measurement results.

[0021] (4) The method is highly versatile: The method provided by this invention is based on the core idea of ​​“establishing a high-precision benchmark model and avoiding local measurement defects through overall registration”. It is universal and applicable not only to thin-walled sleeves, but also, in principle, can be extended to the three-dimensional detection of any complex workpiece with stable macroscopic geometric features, but whose key test area has poor point cloud quality due to structural, material or process reasons. It demonstrates strong technical extensibility.

[0022] (5) Reduced overall costs and support for digital management: By adopting the technical route of CAD models, the reliance on and loss of expensive physical standard parts can be reduced, and non-contact measurement avoids workpiece damage. The digital model and data of the whole process also provide a complete data foundation for quality traceability, process analysis and digital twin applications, which is in line with the development direction of intelligent manufacturing. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0024] Figure 1This is a schematic flowchart of the steps of a thin-walled sleeve measurement method based on calibration and registration in one embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0026] To overcome the shortcomings of existing technologies, this invention provides a measurement method based on digital reference model registration. Specifically, it designs a thin-walled sleeve measurement method based on a two-stage (calibration and registration) framework: decoupling the establishment of an offline calibration model for a high-precision reference from the online rapid measurement registration. In the offline stage, a digital calibration model containing ideal reference geometric features is constructed by scanning a high-precision physical standard part or directly referencing a CAD design model. In the online stage, a robust registration algorithm aligns the scanned point cloud of the workpiece to be measured to the reference coordinate system of this calibration model, and then feature extraction and calculation are performed on the known high-precision reference surface. This method fundamentally isolates the influence of local point cloud quality on the final accuracy during online measurement.

[0027] In one embodiment, such as Figure 1 As shown, the present invention provides a method for measuring thin-walled sleeves based on calibration and registration, the method comprising: Step S100: Perform offline calibration, establish a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establish the spatial coordinate system of the calibration model as the reference coordinate system; the calibration model includes at least a high-precision reference plane and reference circle feature defined by the upper surface of the thin-walled sleeve. Step S200: Perform online registration, including acquiring the measurement point cloud of the thin-walled sleeve to be measured, and registering the measurement point cloud to the reference coordinate system by determining the spatial transformation relationship of transforming the measurement point cloud to the reference coordinate system. Step S300: Project the points corresponding to the upper surface in the measurement point cloud registered in the reference coordinate system onto the reference plane; perform circle fitting on the projected points on the reference plane to obtain the fitting circle parameters in the reference coordinate system. Step S400: Based on the spatial transformation relationship, the center coordinates of the fitted circle parameters are inversely transformed to the original coordinate system where the measured point cloud is located, and the measurement results are output.

[0028] In one embodiment, a calibration model is established by measuring thin-walled sleeves based on scanning physical standard parts, using a thin-walled sleeve (reference thin-walled sleeve) that has been authoritatively verified as a physical standard part.

[0029] Specifically, in step S100, offline calibration is performed to establish a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and the spatial coordinate system of the calibration model is established as the reference coordinate system. This includes: (1) Data acquisition is performed to obtain the calibration point cloud of the reference thin-walled sleeve.

[0030] A high-precision blue light structured light 3D scanner was used to stably clamp the reference thin-walled sleeve, serving as a physical standard part, onto the worktable to acquire complete point cloud data. Scanner parameters (such as exposure time and light intensity) were adjusted, and the scanner was scanned from multiple angles to ensure complete acquisition of point cloud data for its upper surface, outer cylindrical surface, inner hole wall, and lower end face. The data was then stitched together using software to obtain a complete and intact calibration point cloud. .

[0031] (2) Establish a calibration model based on the calibration point cloud, including high-precision reference plane and reference circle features, wherein the reference circle features include at least the center and radius parameters. Specifically: Perform high-precision benchmark feature extraction: Open the point cloud processing software's display interface that contains... The data file contains the point cloud of the upper surface of the reference thin-walled sleeve; In the software display interface, select multiple feature points (no fewer than 6 feature points) precisely and evenly on the point cloud of the upper surface. These feature points should be distributed as much as possible on the circumference of the theoretical annulus (of the annular end face of the upper surface); Using multiple selected feature points, a plane is fitted using the least squares method. As a high-precision reference plane for subsequent measurements, The plane equation is expressed as: ; in, These are coordinates in a three-dimensional spatial coordinate system. These are the equation parameters of the reference plane; Performing circle (feature) fitting includes using the selected multiple feature points on a high-precision reference plane. Project the image onto the projection point and perform circle fitting using the least squares method to obtain the result. The center of the reference circle on the plane coordinates and calibration radius The reference circle refers to the precise circular geometric feature that represents the ideal contour and position of the upper surface (annular end face) of the thin-walled sleeve that needs to be calibrated, and the calibration model parameters, including at least the calibration circle center and calibration radius, are determined.

[0032] Determine the reference coordinate system and establish the calibration model: To determine the center of the circle With the origin of the coordinate system and the reference plane as the reference point. With the normal vector of Z as the Z-axis, establish a right-hand rectangular coordinate system, which is defined as the reference coordinate system for this measurement. reference plane Equation parameters, calibration center Calibration radius Together with the definition of the reference coordinate system, they are packaged and saved as a structured and digital calibration model. This calibration model will be used for the measurement of all workpieces of the same type.

[0033] Further, in step S200, determining the spatial transformation relationship for registering the measured point cloud to the reference coordinate system includes: Step S210: Using the same 3D scanning equipment and settings as in the offline calibration stage, collect complete point cloud data of the thin-walled sleeve to be tested as the measurement point cloud.

[0034] Specifically, the 3D scanning equipment, lens, and clamping scheme are kept completely consistent with those used in the calibration stage. Data acquisition of the test piece is performed by scanning the thin-walled sleeve under test to obtain the measurement point cloud. .

[0035] Step S220, the measured point cloud With calibration point cloud Perform 3D point cloud registration and calculate the optimal spatial transformation matrix between the measured point cloud and the calibration point cloud in the reference coordinate system, including: (1) Perform global coarse registration based on Fast Point Feature Histograms (FPFH) features: The measurement point cloud was measured respectively. With calibration point cloud Keypoint sampling is performed (e.g., using ISS, SIFT3D, or uniform sampling algorithms) to obtain keypoint subsets of two types of point clouds. and This approach significantly reduces the amount of subsequent calculations while preserving the workpiece's distinctive geometric features. For key point subset and For each keypoint in the dataset, compute the FPFH feature descriptor: for each keypoint Calculate its radius as The normal vectors of all points in the neighborhood; calculate With each of its neighboring points The three angular eigenvalues ​​between , used to describe and The spatial relationship between the normal vector and the line connecting the two points; the angular feature values ​​are statistically analyzed into a multidimensional histogram, and... The weighted sum of all neighboring points is used to obtain the final point. Generate a fixed-length vector, which is the key point. The FPFH feature descriptor is used to digitally characterize the geometry of the local surface around the keypoint; Perform feature matching and output coarse registration transformation matrix: for keypoint subsets Each key point in The process involves finding the points most similar to the FPFH feature descriptors (e.g., with the smallest Euclidean distance) to form an initial set of matched point pairs. Due to noise and repetitive structures, the initial matching contains a large number of mismatches. The Random Sampling Consensus (RANSAC) algorithm is used for robustness estimation and mismatch removal. Finally, the set of reliable matched point pairs after mismatch removal and the coarse registration transformation matrix are output. .

[0036] The robustness estimation and mismatch removal using the Random Sampling Consensus (RANSAC) algorithm includes: Randomly select 3 matching point pairs from the initial set of matching point pairs (to solve for the minimum set of rigid body transformations); A candidate is calculated from these 3 matching point pairs. Rotation matrix and Translation vector ; The rotation matrix Translation vector A deterministic transformation is applied to a subset of key points. Calculate how many pairs of matching points have a distance less than a threshold after transformation. (That is, interior point); Repeat the above process for a preset number of iterations (e.g., 1000 times), and retain the optimal transformation with the largest number of interior points whose distance is less than a threshold. In other words, matching points in the original set of matching point pairs whose distance is not less than the threshold have been removed (false matches are eliminated), resulting in a set of reliable matching point pairs, in which the subset of key points is obtained. The set of point clouds in the data is denoted as . , falling The set of point clouds in the data is denoted as . .

[0037] Utilizing optimal transformation , Combination (As the last row in the matrix) Divide into blocks to form a homogeneous matrix, as coarse registration transformation matrix .

[0038] (2) Perform precise registration based on the Iterative Closest Point (ICP) algorithm.

[0039] Iterative optimization is performed using a variant of the Point-to-Plane (ICP) algorithm, with each (the...)... The iteration process (number of iterations) includes: for Each point in the calibration point cloud Find the nearest point in the middle and obtain the surface normal vector at that nearest point; By minimizing Point to The sum of squared distances to the tangent plane at the nearest point (which is the distance from a point to the surface, not the distance from a point to a point) is used to construct the objective function of the linear least squares problem, so as to guide the point cloud to slide along the surface normal, resulting in faster convergence and higher accuracy. By solving the above linear least squares problem, an optimal incremental transformation is calculated: the rotation matrix. Translation vector ; Update the current point cloud using the incremental transformation described above: ; Using the and Construct the corresponding incremental transformation matrix in blocks. (and coarse registration transformation matrix) (Same construction method) Using the incremental transformation matrix Update the cumulative transformation results : ; The iteration stops when the change in root mean square error (RMSE) between two consecutive iterations is less than a preset iteration threshold (e.g., 0.001 mm) or when the preset maximum number of iterations is reached. Output the optimal spatial transformation matrix to characterize the spatial transformation relationship: .

[0040] Using the optimal spatial transformation matrix Measure point cloud Register and align to the reference coordinate system to obtain the registered measurement point cloud. .

[0041] Further, in step S300, the points corresponding to the upper surface in the measurement point cloud registered to the reference coordinate system are projected onto the reference plane; circle fitting is performed on the projected points on the reference plane to obtain the fitted circle parameters in the reference coordinate system. This includes: Registered measurement point cloud Points belonging to the upper surface are vertically projected onto the calibration model. reference plane Above, due to It is precisely known, and this projection operation does not introduce fitting error; By performing circle fitting on the projection points on the reference plane, the center of the fitted circle in the reference coordinate system is obtained. and fitted radius Because the projection datum is accurate and the point cloud is aligned with the calibration model, the fitted circle center is... and fitted radius It is the precise value in the reference coordinate system.

[0042] Finally, in step S400, according to the spatial transformation relationship, the center coordinates of the fitted circle parameters are inversely transformed to the original coordinate system where the measured point cloud is located, and the measurement results are output.

[0043] Specifically, using the optimal spatial transformation matrix The inverse matrix (inverse transformation) , fit the center of the circle Inverse transform to measurement point cloud In the original coordinate system in which it is located, the coordinates of the measured center of the circle are used as the coordinates. And by calculating the fitting radius The error between the measured radius and the calibration radius is used to determine the accuracy of the error.

[0044] The circle fitting is preferably performed using the least squares method, but other methods can also be used, such as algebraic fitting, geometric fitting, Pratt Fit, Taubin Fit, etc.

[0045] In one embodiment, the high-precision blue light structured light 3D scanner uses GOM ATOS Q, and the point cloud processing software uses Geomagic Control X. During high-precision benchmark feature extraction, eight feature points are extracted from the upper surface point cloud; corresponding to the acquired calibration point cloud. Its point cloud density is not less than 0.02 mm, and the overall noise level is less than 0.005 mm; when the least squares method is used to fit the reference plane, the fitting residual is less than 0.001 mm.

[0046] In one embodiment, a contact probe (e.g., a high-precision coordinate measuring machine) is used to acquire measurement data to establish a calibration model. .

[0047] Specifically, in step S100, offline calibration is performed to establish a calibration model that includes the geometric features of the thin-walled sleeve theoretical design. And establish the calibration model. The spatial coordinate system is the reference coordinate system. It includes: (1) The upper surface of the reference thin-walled sleeve is measured using a contact probe to obtain measurement data.

[0048] The reference thin-walled sleeve, which serves as a physical standard part, is placed on a high-precision coordinate measuring machine (CMM) for data acquisition. Then, the contact probe of the CMM is used to directly and accurately measure the three-dimensional coordinates of multiple points on the upper surface.

[0049] (2) The calibration model parameters of the high-precision reference plane and reference circle features are obtained directly based on the measurement data fitting. The calibration model parameters include at least the equation parameters of the reference plane and the calibration center and calibration radius of the reference circle. The flatness of the upper surface is obtained during the measurement process to verify the accuracy of the reference plane.

[0050] A high-precision reference plane is fitted using CMM software. The equation parameters are obtained; by measuring multiple points on the annular profile of the upper surface, the reference circle is directly fitted by CMM software to obtain the calibration circle center. and calibration radius These calibration model parameters (equation parameters of the reference plane, calibration center, and calibration radius) are usually output directly in the form of a mathematical model.

[0051] During the measurement process, the flatness of the upper surface can also be obtained as an evaluation index for fitting and obtaining the features of the reference plane and reference circle. This is used to judge the accuracy of the features of the reference plane and reference circle, ensuring that the upper surface of the physical standard part (reference thin-walled sleeve) is flat enough to qualify as a high-precision reference.

[0052] Furthermore, in step S200, online registration is performed, including: The measurement point cloud of the thin-walled sleeve under test is still obtained using a 3D scanning device (such as GOM ATOS Q). ; Since the measuring device and the calibration device (contact probe) are different in this embodiment, and there is no calibration point cloud, during registration, the measuring point cloud is directly compared with the calibration model established in the previous step S100. The geometric features are used for registration and alignment, including using geometric feature-based registration methods or the Iterative Closest Point (ICP) algorithm based on the point-to-calibration model to calculate the optimal spatial transformation matrix.

[0053] The geometric feature-based registration method includes: from the measured point cloud Features such as the plane and the axis of the cylinder are fitted and compared with... Known reference plane By matching geometric features such as theoretical axes, the optimal spatial transformation matrix can be directly calculated. .

[0054] The ICP algorithm based on the point-to-calibration model includes: measuring the point cloud... Points and calibration model CAD surfaces (which can be derived from calibration models) (using simple parameter generation) iterative nearest point search and alignment, and calculate the optimal spatial transformation matrix. .

[0055] Subsequent steps S300 and S400 are the same as in the previous embodiment.

[0056] In another embodiment of the invention, a CAD design model is directly used as the calibration model for measurement reference. No physical standard parts are required, and therefore no equipment is needed to collect and calibrate point clouds.

[0057] Specifically, in step S100, offline calibration is performed to establish a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and the spatial coordinate system of the calibration model is established as the reference coordinate system, including: The CAD design model of the thin-walled sleeve to be tested is directly used as the calibration model for the theoretical design geometric features of the thin-walled sleeve: Obtain the accurate CAD design model of the thin-walled sleeve to be tested. The original contains an ideal upper surface plane. Center and radius ; CAD design model As a digital calibration model , will the As a reference plane center and radius Each as a calibration model The calibration center and calibration radius of the reference circle; Establishing the spatial coordinate system of the calibration model as the reference coordinate system includes: By scanning a reference thin-walled sleeve of known dimensions, a fixed spatial transformation relationship between the coordinate system of the three-dimensional scanning device and the coordinate system of the CAD design model is calculated and determined, and the coordinate system of the CAD design model is established as the reference coordinate system.

[0058] Specifically, a 3D scanning device is used to scan a known and universally applicable reference thin-walled sleeve (e.g., a standard sphere array); by aligning the scanned point cloud of the reference thin-walled sleeve with its corresponding CAD design model, the coordinate system of the 3D scanning device to the CAD design model is accurately calculated. Fixed spatial transformation matrix between coordinate systems ; through the fixed spatial transformation matrix The coordinate system of the CAD design model is established as the reference coordinate system for measurement.

[0059] Further, in step S200, the measurement point cloud of the thin-walled sleeve to be measured is obtained, and the measurement point cloud is transformed and registered to the reference coordinate system by determining the spatial transformation relationship between the measurement point cloud and the reference coordinate system, including: The measurement point cloud of the thin-walled sleeve under test is still obtained using a 3D scanning device (such as GOM ATOS Q). ; The measured point cloud is transformed to the reference coordinate system using the fixed spatial transformation relationship: directly using the fixed spatial transformation matrix. Measure point cloud Transform to the reference coordinate system of the CAD design model: ; In the specific calculation, substitute the coordinates of the specific points in the measured point cloud; By transforming the measurement point cloud to a reference coordinate system The model-based iterative nearest-point registration is performed with the surface of the CAD design model, and the corresponding correction transformation is calculated. To compensate for minor differences in clamping operations, a fine registration is performed, including calling the surface geometry information of the CAD model and executing the point-to-CAD design model ICP algorithm. In each iteration, the algorithm calculates... Point to (that is) The nearest projection point of the model surface is used as the objective function to optimize and solve a small correction transformation by minimizing the distance between these points. ; Combining the fixed spatial transformation relationship with the corrected transformation yields the spatial transformation relationship between the measured point cloud and the reference coordinate system, describing the spatial transformation relationship from the measured point cloud. The optimal spatial transformation matrix for the spatial transformation relationship between the reference coordinate system and the reference coordinate system is: .

[0060] Subsequent steps S300 and S400 are the same as in the previous embodiment.

[0061] In summary, the beneficial effects obtained by the calibration-registration-based thin-walled sleeve measurement method provided by the present invention through the above embodiments include: (1) Significantly improved measurement accuracy: By registering to the calibration model (whether from fitting high-standard parts or directly referencing CAD models), the initial plane fitting error is completely avoided, and the high-precision known plane is directly used. As a projection reference, it eliminates the reference surface error caused by poor point cloud quality on the upper surface from the root.

[0062] (2) Improved stability and repeatability of the measurement system: Once the calibration model M is established in the offline stage, its geometric features (plane, center) as the reference are fixed and highly accurate, and can be used repeatedly. The consistency of subsequent measurements mainly depends on the accuracy of point cloud registration, and the registration algorithm utilizes the overall and rich geometric features of the part (such as inner and outer cylindrical surfaces), and its stability and repeatability are far higher than the traditional method of fitting based only on a narrow upper surface area.

[0063] (3) Balancing measurement accuracy and online efficiency: This invention places the most critical task of "establishing a high-precision benchmark" in a one-time offline calibration stage, allowing for more time and manpower to ensure optimal results. In the online measurement stage, the process is fully automated (automatic acquisition, registration, calculation, and output), with single-piece measurement efficiency comparable to or even higher than traditional methods, while achieving a qualitative leap in the accuracy and reliability of the measurement results.

[0064] (4) The method is highly versatile: The method provided by this invention is based on the core idea of ​​“establishing a high-precision benchmark model and avoiding local measurement defects through overall registration”. It is universal and applicable not only to thin-walled sleeves, but also, in principle, can be extended to the three-dimensional detection of any complex workpiece with stable macroscopic geometric features, but whose key test area has poor point cloud quality due to structural, material or process reasons. It demonstrates strong technical extensibility.

[0065] (5) Reduced overall costs and support for digital management: By adopting the technical route of CAD models, the reliance on and loss of expensive physical standard parts can be reduced, and non-contact measurement avoids workpiece damage. The digital model and data of the whole process also provide a complete data foundation for quality traceability, process analysis and digital twin applications, which is in line with the development direction of intelligent manufacturing.

[0066] In one embodiment, the present invention also provides a thin-walled sleeve measuring device based on calibration registration, comprising: The first module is used for offline calibration, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establishing the spatial coordinate system of the calibration model as the reference coordinate system; the calibration model includes at least a high-precision reference plane and reference circle feature defined by the upper surface of the thin-walled sleeve. The second module is used for online registration, including acquiring the measurement point cloud of the thin-walled sleeve to be measured, and registering the measurement point cloud to the reference coordinate system by determining the spatial transformation relationship of transforming the measurement point cloud to the reference coordinate system. The third module is used to project the points corresponding to the upper surface in the measurement point cloud registered in the reference coordinate system onto the reference plane; and to perform circle fitting on the projected points on the reference plane to obtain the fitting circle parameters in the reference coordinate system. The fourth module is used to inversely transform the center coordinates of the fitted circle parameters to the original coordinate system of the measured point cloud according to the spatial transformation relationship, and output the measurement results.

[0067] On the other hand, in one embodiment, the present invention provides a computer device including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the calibration-registration-based thin-walled sleeve measurement method provided in any of the above embodiments. The computer device may be a server. The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store sample data. The network interface of the computer device is used for communication with external terminals via a network connection.

[0068] On the other hand, in one embodiment of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the calibration-registration-based thin-walled sleeve measurement method provided in any of the above embodiments.

[0069] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0070] Matters not covered in this invention are common knowledge.

[0071] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0072] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

[0073] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for measuring thin-walled sleeves based on calibration and registration, characterized in that, include: Step S100: Perform offline calibration, establish a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establish the spatial coordinate system of the calibration model as the reference coordinate system; the calibration model includes at least a high-precision reference plane and reference circle feature defined by the upper surface of the thin-walled sleeve. Step S200: Perform online registration, including acquiring the measurement point cloud of the thin-walled sleeve to be measured, and registering the measurement point cloud to the reference coordinate system by determining the spatial transformation relationship of transforming the measurement point cloud to the reference coordinate system. Step S300: Project the points in the measurement point cloud that correspond to the upper surface, which are registered in the reference coordinate system, onto the reference plane; The projected points are fitted with a circle on the reference plane to obtain the fitted circle parameters in the reference coordinate system. Step S400: Based on the spatial transformation relationship, the center coordinates of the fitted circle parameters are inversely transformed to the original coordinate system where the measured point cloud is located, and the measurement results are output.

2. The method for measuring thin-walled sleeves based on calibration registration according to claim 1, characterized in that, In step S100, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design includes: Obtaining the calibration point cloud of the reference thin-walled sleeve includes using a 3D scanning device to collect complete point cloud data of a known qualified thin-walled sleeve as a reference, which is used as the calibration point cloud; the calibration point cloud completely covers the point cloud data of the upper surface, outer cylindrical surface, inner hole wall and lower end surface of the reference thin-walled sleeve. A calibration model containing high-precision reference plane and reference circle features is established based on the calibration point cloud.

3. The method for measuring thin-walled sleeves based on calibration and registration according to claim 2, characterized in that, The process of establishing a calibration model based on the calibration point cloud, including high-precision features of the reference plane and reference circle, includes: In the upper surface point cloud of the calibration point cloud, multiple feature points are uniformly selected; Based on the selected feature points, the high-precision reference plane and reference circle features are fitted using the least squares method to obtain calibration model parameters that include at least the calibration circle center and calibration radius.

4. The method for measuring thin-walled sleeves based on calibration and registration according to claim 1, characterized in that, In step S100, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design includes: The upper surface of the reference thin-walled sleeve was measured using a contact probe to obtain measurement data; The calibration model parameters of the high-precision reference plane and reference circle features are obtained directly based on the measurement data fitting; the calibration model parameters include at least the equation parameters of the reference plane and the calibration center and calibration radius of the reference circle, and the flatness of the upper surface is obtained during the measurement process to verify the accuracy of the reference plane.

5. The method for measuring thin-walled sleeves based on calibration and registration according to claim 1, characterized in that, In step S100, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establishing the spatial coordinate system of the calibration model as the reference coordinate system, includes: The CAD design model of the thin-walled sleeve to be tested is directly used as the calibration model for the theoretical design geometric features of the thin-walled sleeve; Establishing the spatial coordinate system of the calibration model as the reference coordinate system includes: By scanning a reference thin-walled sleeve of known dimensions, a fixed spatial transformation relationship between the coordinate system of the three-dimensional scanning device and the coordinate system of the CAD design model is calculated and determined, and the coordinate system of the CAD design model is established as the reference coordinate system.

6. The method for measuring thin-walled sleeves based on calibration registration according to claim 3 or 4, characterized in that, In step S200, determining the spatial transformation relationship for registering the measured point cloud to the reference coordinate system includes: Using the same 3D scanning equipment and settings as in the offline calibration phase, complete point cloud data of the thin-walled sleeve under test was acquired as the measurement point cloud. The measured point cloud and the calibration point cloud are registered in three dimensions, and the optimal spatial transformation matrix between the measured point cloud and the calibration point cloud in the reference coordinate system is calculated.

7. The method for measuring thin-walled sleeves based on calibration and registration according to claim 5, characterized in that, In step S200, determining the spatial transformation relationship for registering the measured point cloud to the reference coordinate system includes: The measured point cloud is transformed to a reference coordinate system using the fixed spatial transformation relationship. By performing model-based iterative nearest-point registration between the measured point cloud transformed to the reference coordinate system and the surface of the CAD design model, the corresponding correction transformation is calculated. By combining the fixed spatial transformation relationship with the corrected transformation, the spatial transformation relationship between the measured point cloud and the reference coordinate system is obtained.

8. The method for measuring thin-walled sleeves based on calibration registration according to claim 1, characterized in that, In step S300, the circle fitting is performed using the least squares method.

9. A thin-walled sleeve measuring device based on calibration and registration, characterized in that, include: The first module is used for offline calibration, establishing a calibration model that includes the geometric features of the thin-walled sleeve theoretical design, and establishing the spatial coordinate system of the calibration model as the reference coordinate system; the calibration model includes at least a high-precision reference plane and reference circle feature defined by the upper surface of the thin-walled sleeve. The second module is used for online registration, including acquiring the measurement point cloud of the thin-walled sleeve to be measured, and registering the measurement point cloud to the reference coordinate system by determining the spatial transformation relationship of transforming the measurement point cloud to the reference coordinate system. The third module is used to project the points corresponding to the upper surface in the measurement point cloud registered in the reference coordinate system onto the reference plane; The projected points are fitted with a circle on the reference plane to obtain the fitted circle parameters in the reference coordinate system. The fourth module is used to inversely transform the center coordinates of the fitted circle parameters to the original coordinate system of the measured point cloud according to the spatial transformation relationship, and output the measurement results.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the thin-walled sleeve measurement method based on calibration registration as described in claim 1.

Images