A Dynamic Error Compensation System and Method for Three-Dimensional Inspection of High-Speed ​​Components Based on Beam Reference

By using a beam reference method to observe and compensate for the dynamic offset of the 3D profilometer in real time, the problem of dynamic error in the detection of tall components is solved, the detection accuracy and scanning speed are improved, and the dependence on mechanical structure is reduced.

CN123111084APending Publication Date: 2026-11-17HARBIN ENG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611619913.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the inspection of tall components, existing 3D profilometers suffer from problems due to the difficulty in maintaining the rigidity of the inspection slide and its supporting structure. This leads to dynamic offsets in the X and Y directions caused by cantilever deflection, guide rail gaps, and motor vibration, which are superimposed on the point cloud, resulting in lateral misalignment, profile distortion, and dimensional errors. Existing solutions cannot effectively compensate for these dynamic errors in real time.

Method used

By adopting a beam-based reference method, a mapping relationship between the displacement of the center pixel of the light spot image and the offset of the detection slide is established. A high frame rate area array camera is used to observe the light spot trajectory in real time, calculate and reverse compensate for the dynamic offset of the 3D point cloud, and reduce the dependence on the stiffness of the mechanical guide rail.

Benefits of technology

It enables the acquisition of high-precision 3D point clouds even with slight vibrations in the mechanical system, reduces the dependence on heavy, high-rigidity mechanical structures, improves scanning speed adaptability and detection accuracy, and reduces lateral misalignment and dimensional errors.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123111084A_ABST
    Figure CN123111084A_ABST
Patent Text Reader

Abstract

This invention discloses a dynamic error compensation system and method for three-dimensional inspection of high-structure components based on a beam reference. It belongs to the fields of three-dimensional contour inspection, precision measurement, and error compensation. The system includes a three-dimensional profilometer, an inspection slide, a beam reference module, a high frame rate area array camera, and a synchronization and compensation processing module. The beam reference module is independent of the inspection slide and fixed to a stationary base, emitting a stable reference beam. The high frame rate area array camera and the three-dimensional profilometer are rigidly mounted on the same inspection slide and move up and down with the slide. The reference beam directly enters the lens to form a light spot. The synchronization and compensation processing module extracts the center of the light spot and generates a trajectory. It matches the three-dimensional contour data with the light spot image according to the sampling time, frame number, or Z-axis position, calculates the dynamic offset of the inspection slide and the three-dimensional profilometer in the X and Y directions, and compensates for the reverse coordinates of the original three-dimensional point cloud. This reduces the dependence of inspection accuracy on the absolute stiffness of the guide rail and cantilever, and suppresses lateral misalignment and contour distortion of the point cloud.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the fields of three-dimensional contour detection, precision measurement and error compensation technology, and in particular to a system and method for real-time observation of the dynamic offset of the detection head and reverse coordinate compensation of the three-dimensional point cloud during the three-dimensional scanning detection of tall components. Background Technology

[0002] 3D profilometers are widely used in scenarios such as component shape inspection, dimensional measurement, surface defect identification, and reverse modeling. For components with considerable height, a common inspection method is to mount the 3D profilometer on a sliding table that can be raised and lowered along the Z-axis. The sliding table then drives the profilometer to scan the component from top to bottom or bottom to top, thereby obtaining a complete 3D point cloud.

[0003] However, in actual testing, it is difficult to maintain ideal rigidity in the testing slide and its supporting structure. Especially when the 3D profilometer is installed at the end of the cantilever, as the Z-axis position changes, cantilever deflection, guide rail clearance, lead screw or linear motor vibration, assembly eccentricity, and acceleration / deceleration impacts can cause the profilometer to produce micron-level or even millimeter-level physical sway in the X and Y directions. This sway is directly superimposed on the point cloud of the component being measured, causing lateral misalignment of the point cloud, profile distortion, and dimensional errors.

[0004] Existing solutions typically improve detection accuracy by increasing guide rail stiffness, adding mechanical supports, reducing scanning speed, or performing offline calibration. These methods suffer from high cost, bulky structures, poor adaptability, or inability to reflect real-time dynamic errors. If only the mechanical spatial synchronization between the 3D profilometer and the detection slide is relied upon, assuming the motion mechanism maintains ideal linear motion during scanning, it becomes difficult to compensate for transient jitter and attitude drift during actual motion.

[0005] Therefore, it is necessary to propose a detection system that does not rely solely on the absolute stiffness of the mechanical guide rail as the sole source of accuracy. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide a dynamic error compensation system and method for three-dimensional detection of high-rise components based on a beam reference. This system addresses how to observe and compensate for the dynamic offset of the three-dimensional profilometer in the X and Y directions caused by cantilever deflection, guide rail clearance, motor vibration, assembly eccentricity, and acceleration / deceleration impact when the detection slide moves up and down along the Z-axis during the three-dimensional scanning and detection of high-rise components. This avoids the dynamic offset being directly superimposed on the three-dimensional point cloud, causing lateral misalignment, contour distortion, and dimensional errors. In other words, it reduces the dependence of detection accuracy on the absolute stiffness of the mechanical guide rail and cantilever structure.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A dynamic error compensation method for three-dimensional detection of high-structure components based on a beam reference includes: S1. Establish the Z-axis position reference and beam reference, and calibrate the mapping relationship between the center pixel displacement of the spot image and the actual offset of the detection slide and the three-dimensional profilometer in the X and Y directions. S2, obtain the Z-axis position, and synchronously trigger the 3D profilometer and high frame rate area array camera according to the preset displacement, and establish a correspondence between the original 3D point cloud and the spot image according to at least one of the sampling time, frame number or Z-axis position. S3 drives the detection slide to scan the component being measured along the Z-axis, simultaneously acquiring the original three-dimensional point cloud and spot images; S4, extract the center coordinates of the light spot to generate the motion trajectory, and calculate the dynamic offset of the detection slide and the three-dimensional profilometer in the X and Y directions; S5, Match the dynamic offset to the original 3D point cloud according to the corresponding relationship; S6 performs reverse coordinate compensation on the original 3D point cloud to obtain the compensated 3D point cloud.

[0008] Therefore, the method system of this invention should introduce a stationary beam reference decoupled from the mechanical motion system, and optionally utilize the actual Z-axis position to improve the consistency of sampling and data matching. This allows the area array camera moving with the detection slide to directly receive the incident beam and observe the beam trajectory, thereby performing reverse compensation on the 3D point cloud at the algorithm level. This solves the problem that the dynamic offsets in the X and Y directions caused by cantilever effect, motor vibration, guide rail clearance, and structural flexibility when the detection slide moves along the Z-axis are difficult to observe and compensate for in real time by conventional 3D contour scanning systems.

[0009] In the above technical solution, in S1, the beam reference module is fixed on a stationary base independent of the detection slide, and the three-dimensional profilometer and the high frame rate area array camera are installed on the same detection slide.

[0010] In the above technical solution, the specific calibration in S1 involves establishing a coordinate mapping between the motion of the light spot image plane and the actual spatial motion of the instrument. The mapping relationship can be a scaling factor relationship, an affine mapping relationship, or a homography mapping relationship.

[0011] That is: the light spot moved on the camera's imaging plane. Each pixel corresponds to the actual movement in millimeters of the detection slide and 3D profilometer in the X and Y directions. Specifically, it calibrates three layers of relationships: Coordinate reference correspondence: Image plane pixel coordinate system The spatial coordinate system of the detection slide / 3D profilometer The axial correspondence includes whether they are in the same direction, whether they are mirror images, and whether there is inter-axis rotation; Scale ratio: How much actual displacement corresponds to each pixel, i.e. , Units such as mm / pixel; Mapping model: When the camera optical axis is not perfectly perpendicular to the reference beam, or when there is mounting tilt or perspective, it is described using a homography matrix H or an affine matrix. .

[0012] In the above technical solution, in S1, the beam reference module includes a laser collimator, a laser collimator, or a fixed beam generating device, and the reference beam is a vertical beam or a collimated beam.

[0013] In the above technical solution, the reference beam is preferably configured to control light intensity and suppress stray light by passing through a filter, attenuator, light limiting hole or light shield.

[0014] In the above technical solution, in S4, the center coordinates of the light spots in 1000 or more consecutive light spot images are located at the sub-pixel level, and the dynamic offset curve is obtained by moving average filtering, median filtering, Kalman filtering or spline fitting.

[0015] In the above technical solution, the center coordinates of the light spot in S4 are determined by at least one of the following methods: region of interest clipping, grayscale threshold segmentation, connected component filtering, centroid method, ellipse fitting, or Gaussian fitting.

[0016] In the above technical solution, in S6, when only the translational deviations in the X and Y directions are calculated, two-dimensional lateral translation reverse compensation is performed on the original three-dimensional point cloud; when the translational deviations and angular deviations are calculated, a rigid body reverse coordinate transformation including translations in the X, Y, and Z directions and rotations around at least one coordinate axis is performed on the original three-dimensional point cloud.

[0017] In the above technical solution, in S2, the Z-axis position is determined by the Z-axis position reference module according to position counting, and the original three-dimensional point cloud and spot image are synchronously acquired through the hardware equidistant trigger module.

[0018] Preferably, the Z-axis position reference module is a magnetic scale, optical scale, encoder, or linear displacement measuring device.

[0019] In the above technical solution, S1 further includes: acquiring a light spot image in a static state before detection to determine the initial reference position.

[0020] In S4, the real-time light spot center coordinates are compared with the initial reference position to obtain the dynamic offset of the detection slide and the three-dimensional profilometer in the X and Y directions.

[0021] The core of this invention is not a pure spatial synchronization mechanism or a positional equidistant triggering mechanism, but rather the setting of a stationary beam reference independent of the mechanical kinematic chain. The motion trajectory of the reference beam is continuously observed by an area array camera that moves together with the three-dimensional profilometer. The dynamic offset of the sliding stage is calculated and detected, and the dynamic offset is converted into a reverse compensation amount of three-dimensional point cloud coordinates.

[0022] The present invention also provides a dynamic error compensation system for three-dimensional detection of high-structure components based on a beam reference, comprising: A 3D profilometer is used to scan the height of a component being measured and output the original 3D point cloud. The detection slide moves the 3D profilometer up and down along the Z-axis. The beam reference module, independent of the detection slide and fixed to the stationary base, is used to emit a stable reference beam; A high frame rate area array camera is mounted on the same detection slide as the 3D profilometer and moves with the detection slide. The lens is oriented toward the reference beam so that the reference beam is incident and imaged to form a light spot. The synchronization and compensation processing module is used to match the original 3D point cloud with the spot image according to at least one of the sampling time, frame number or Z-axis position. It calculates the dynamic offset of the detection slide and 3D profilometer in the X and Y directions by the change of spot position, and compensates the original 3D point cloud coordinates in reverse to obtain the compensated 3D point cloud.

[0023] In the above technical solution, the beam reference module includes a laser collimator, a laser collimator, or a fixed beam generating device. The reference beam is a vertical beam or a collimated beam, and its light intensity is controlled and stray light is suppressed by a filter, an attenuator, a light limiting hole, or a light shield.

[0024] The above technical solution also includes a Z-axis position reference module and a hardware equidistant trigger module; the Z-axis position reference module is a magnetic grating ruler, optical grating ruler, encoder or linear displacement measuring device, and the hardware equidistant trigger module outputs a trigger signal according to position counting, so that the Z coordinate of the original three-dimensional point cloud is determined by the Z-axis position reference module.

[0025] In the above technical solution, the synchronization and compensation processing module includes a calibration unit, which is used to establish the transformation relationship between the high frame rate area array camera pixel coordinate system, the three-dimensional profilometer coordinate system, the detection slide coordinate system and the measured height component coordinate system, and to determine the proportional coefficient, affine mapping relationship or homography mapping relationship between the spot pixel displacement and the actual offset of the detection slide and the three-dimensional profilometer in the X and Y directions.

[0026] In the above technical solution, the synchronization and compensation processing module includes a light spot positioning unit, which determines the sub-pixel center coordinates of the light spot by at least one of the following methods: region of interest cropping, grayscale threshold segmentation, connected component filtering, centroid method, ellipse fitting or Gaussian fitting.

[0027] In the above technical solution, the synchronization and compensation processing module includes a trajectory analysis unit, which is used to perform trajectory fitting, filtering, outlier removal and error curve generation on the center coordinates of multiple consecutive light spots to obtain the dynamic offset curve in the Z-axis lifting stroke.

[0028] In the above technical solution, the reverse coordinate compensation is: for points in the original 3D point cloud Based on the corresponding dynamic offset , , revised to When calculating the angular deviation, a rigid body reverse coordinate transformation is performed based on the translational deviation and the angular deviation.

[0029] In the above technical solution, the beam reference module provides at least two reference light points, or the high frame rate area array camera is at least two; the synchronization and compensation processing module calculates translational deviation and angular deviation based on the imaging results of multiple reference light points or multiple high frame rate area array cameras, and performs rigid body inverse coordinate transformation including translation and rotation.

[0030] The beam reference module and the mechanical motion chain of the detection slide are independent of each other and are fixed to the foundation, independent reference base or external stabilizing support; the rigid reference plate of the detection head is installed on the detection slide and is driven by the detection slide to move up and down along the Z-axis; the three-dimensional profilometer and the high frame rate area array camera are rigidly installed on the same rigid reference plate of the detection head; the lens of the high frame rate area array camera faces the beam reference module, so that the stable reference beam is directly incident on the high frame rate area array camera and forms a light spot on the imaging surface.

[0031] The system can optionally use position counting and equidistant triggering methods to improve longitudinal sampling consistency, but its core lies in using an independent beam reference to observe the dynamic offset of the mechanical motion chain in real time and perform reverse compensation on the three-dimensional point cloud at the algorithm level, thereby reducing the dependence of detection accuracy on the absolute stiffness of the mechanical guide rail and cantilever structure.

[0032] The working principle of this invention is as follows: The core of this invention is not a pure spatial synchronization mechanism or a positional equidistant triggering mechanism, but rather the setting of a stationary beam reference independent of the mechanical kinematic chain. The system uses an area array camera that moves together with the 3D profilometer to continuously observe the trajectory of the reference beam, calculates the dynamic offset of the detection slide before detection, and calibrates the pixel coordinate system of the area array camera, the coordinate system of the 3D profilometer, the coordinate system of the detection slide, and the coordinate system of the component being measured to establish a mapping relationship between the pixel displacement of the imaging surface beam and the actual lateral offset of the 3D profilometer.

[0033] During the detection process, the synchronization and compensation processing module calculates the dynamic offsets Δx and Δy of the 3D profilometer in the X and Y directions based on the change of the light spot on the imaging surface relative to the initial reference position, and matches the corresponding offsets to each frame point cloud, each scan line, or each point cloud point according to the frame number or Z-axis position.

[0034] For implementation methods that only require compensation for lateral translation errors, the following compensation relationship can be adopted: ,or ,in For the original point cloud points, For the compensated point cloud points, Sampling time, This corresponds to the height position.

[0035] For implementations requiring compensation for attitude angle errors, multiple reference light points, multiple reference target surfaces, or multiple area array cameras can be set. The translational and rotational deviations of the 3D profilometer are calculated by the relative position changes of multiple reference feature points, and a rigid body inverse coordinate transformation is performed on the original point cloud. , where T(t) represents the dynamic pose transformation matrix of the 3D profilometer relative to the reference state at scanning time t.

[0036] Therefore, this invention discloses a dynamic error compensation system and method for three-dimensional detection of high-structure components based on a beam reference, belonging to the field of three-dimensional contour detection and precision measurement technology. The system includes a three-dimensional profilometer, a detection slide, a beam reference module, a high frame rate area array camera, and a synchronization and compensation processing module. The beam reference module is independently set on the detection slide and fixed on a stationary base, used to emit a stable reference beam; the high frame rate area array camera and the three-dimensional profilometer are fixedly mounted on the same rigid reference plate of the detection head of the detection slide and move synchronously with the detection slide. The reference beam directly enters the lens of the area array camera and forms a light spot on the imaging surface. The synchronization and compensation processing module extracts the center coordinates of the light spot and generates the light spot motion trajectory. Based on the calibration relationship between pixel displacement and physical displacement, it calculates the dynamic offset of the detection head and the three-dimensional profilometer in the X and Y directions, and matches the dynamic offset to the original three-dimensional point cloud according to the sampling time, frame number, or Z-axis position, performing reverse coordinate compensation.

[0037] Compared with the prior art, the present invention has the following beneficial effects: By establishing an independent optical reference frame, the detection accuracy no longer depends entirely on the absolute rigidity of the detection slide, guide rail, and cantilever structure. This allows for the acquisition of high-precision 3D point clouds even when there are minor vibrations in the mechanical system. This reduces the reliance on heavy, high-rigidity mechanical structures for lateral accuracy, which helps to reduce structural weight, lower costs, and improve scanning speed adaptability.

[0038] This invention utilizes a high frame rate area array camera to continuously or trigger the monitoring of the light spot trajectory on the imaging surface. Since the beam reference module is stationary, the position change of the light spot on the imaging surface directly reflects the real dynamic offset of the detection head and the three-dimensional profilometer in the X and Y directions, rather than relying solely on the slide command position or mechanical spatial synchronization relationship. It can obtain the dynamic deviation of the detection slide during the scanning process in real time and can reflect transient errors such as motor vibration, acceleration and deceleration impact, and structural elastic deformation.

[0039] The synchronization and compensation processing module matches the dynamic offset to the original 3D point cloud according to the sampling time, frame number, or Z-axis position and performs reverse coordinate compensation. This can deduct the lateral sway superimposed on the point cloud during the motion from the measurement results, suppress the lateral misalignment, contour distortion, and dimensional error of the point cloud, and improve the 3D reconstruction accuracy of tall components, slender components, or laterally sensitive components.

[0040] Secondly, this invention matches the light spot trajectory with the three-dimensional point cloud frame number or Z-axis position, and performs reverse coordinate compensation on the point cloud at the algorithm level, which can avoid the problem that simple mechanical synchronization or offline static calibration cannot eliminate dynamic errors.

[0041] This invention can utilize 1000 or more frames of spot images to construct an error curve for the entire lifting and lowering stroke, enabling both real-time compensation and high-precision offline correction of the point cloud after detection. It is suitable for detecting tall components, slender components, or parts sensitive to lateral accuracy. Furthermore, it suppresses image noise and occasional misidentification. Attached Figure Description

[0042] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the overall structure of the system of the present invention; Figure 3 This is a schematic diagram of the motion trajectory of the reference light spot in 1000 reference frames; Figure 4 A schematic diagram illustrating the improvement effect of compensating for lateral misalignment of point clouds before and after.

[0043] Explanation of reference numerals in the attached diagram: 1. Component being measured; 2. 3D profilometer; 3. Detection slide; 4. Z-axis lifting mechanism; 5. Area array camera; 6. Laser plumb line; 7. Beam reference; 8. Imaging plane; 9. Synchronization and compensation processing module; 10. Incident light spot; 11. Z-axis motion direction. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0045] Example 1 like Figure 1 and Figure 2As shown, this embodiment provides a dynamic error compensation system for three-dimensional inspection of tall components. The system includes a three-dimensional profilometer 2, an inspection slide 3, a beam reference module (laser plumb line 6), a high frame rate area array camera 5, and a synchronization module. The beam reference module is independent of the inspection slide and fixed to a stationary base, emitting a stable reference beam. The high frame rate area array camera 5 and the three-dimensional profilometer 2 are rigidly mounted on the same inspection head and move up and down with the inspection slide 3. The reference beam or beam reference 7 directly enters the lens to form a light spot. The synchronization and compensation processing module 9 extracts the center of the light spot on the imaging surface 8 and generates a trajectory. It matches the three-dimensional contour data and the light spot image according to the sampling time, frame number, or Z-axis position, calculates the dynamic offset of the inspection head and the three-dimensional profilometer in the X and Y directions, and compensates for the reverse coordinates of the original three-dimensional point cloud. This can reduce the dependence of inspection accuracy on the absolute stiffness of the guide rail and cantilever, and suppress lateral misalignment and contour distortion of the point cloud. The component 1 to be measured is fixed on a separate testing station above a stationary base, independent of the testing slide. The testing slide 3 is mounted on the Z-axis lifting mechanism 4 and driven by the Z-axis lifting mechanism 4 to move up and down along the Z-direction 11. The testing head is mounted on the testing slide 3 and moves up and down synchronously with the testing slide 3. The 3D profilometer 2 and the high frame rate area array camera 5 are mounted together on the same rigid reference plate or testing head. The testing slide is driven by the Z-axis lifting mechanism 4, so that the testing head scans the component 1 to be measured along the height direction of the component 1.

[0046] A beam reference module 6 is installed next to the inspection station. The beam reference module 6 is not installed on the inspection slide 3, nor does it move with the 3D profilometer 2 and the area array camera 5. Instead, it is fixed on the foundation, an independent reference base, or an independent support (on the XY plane) and does not move with the Z-axis, thus providing a static reference beam. The height-to-be-measured component 1 and the beam reference module 6 are arranged at a relative interval on the independent reference base.

[0047] The beam reference module 6 is preferably a laser collimator 6, used to emit a stable vertical reference beam or beam reference 7. The stable beam reference 7 is directly incident on the high frame rate area array camera 5, and the incident light spot 11 forms a light spot on the imaging surface 8 of the area array camera 5.

[0048] The lens of the high frame rate area array camera 5 is oriented towards the beam reference 7, so that the beam reference 7 directly enters the camera lens and forms a light spot on the camera's imaging surface. To avoid overexposure and stray light interference, filters, ND filters, light-limiting holes, light-shielding tubes, or protective windows can be installed at the front of the camera.

[0049] The high frame rate area array camera 5 and the 3D profilometer 2 are fixedly mounted on the same rigid reference plate of the detection head. When the detection head or the rigid reference plate of the detection head wobbles laterally in the X and Y directions, the 3D profilometer 2 and the area array camera 5 shift synchronously as a whole; since the beam reference 7 remains stationary, the change in the position of the light spot on the imaging surface of the area array camera 5 or the trajectory of the light spot can reflect the synchronous lateral shift of the 3D profilometer.

[0050] The synchronization and compensation processing module 9 can be set up independently or integrated into the detection slide or any component. Generally, the synchronization and compensation processing module is set to include a calibration unit, which is used to establish the transformation relationship between the pixel coordinate system of the high frame rate area array camera, the coordinate system of the 3D profilometer, the coordinate system of the detection slide, and the coordinate system of the component being measured, and to determine the scaling factor, affine mapping relationship, or homography mapping relationship between the spot pixel displacement and the actual offset of the detection head and the 3D profilometer in the X and Y directions.

[0051] Before the detection begins, the synchronization and compensation processing module 9 acquires several frames of still reference images and calculates the initial center coordinates of the reference light point. During the testing process, the testing slide moves the 3D profilometer 2 up and down along the Z-axis (Z direction) 11. Due to the cantilever effect, motor vibration, and guide rail clearance, the 3D profilometer 2 and its mounting structure may experience slight wobbling in the X and Y directions.

[0052] Since the beam reference module 6 remains stationary, the relative position of the imaging surface spot in the high frame rate area array camera image changes as the detection head shakes. Therefore, the spot image trajectory can characterize the dynamic offset of the 3D profilometer relative to the stationary beam reference.

[0053] Figure 3 This is a schematic diagram of the movement trajectory of the light spot in the image pixel coordinate system during the measurement process. The horizontal coordinate u represents the pixel column direction, and the vertical coordinate v represents the pixel row direction, with the unit being pixels. The red curve is the movement trajectory of the light spot, and the black dots mark the initial position of the light spot.

[0054] Depend on Figure 3 It is evident that the center of the light spot is not fixed during mechanical movement, but rather drifts irregularly near the initial spot position. The trajectory is roughly distributed within the range of u (approximately 640–1000 pixels) and v (approximately 350–660 pixels), with the offset relative to the initial spot position reaching approximately one hundred pixels. This indicates that the measurement system, while continuously moving in the Z-axis, is accompanied by significant lateral swaying or low-frequency oscillation. This spot drift corresponds to the dynamic lateral offset of the measured component 1 or the 3D profilometer 2 relative to the stationary reference coordinate system, and is one of the reasons for the bending of the original point cloud centerline and the apparent deformation of the contour.

[0055] In this application, the image acquisition device acquires spot images frame by frame, extracts the center coordinates (u, v) of the spot in each frame, and the synchronization and compensation processing module associates the center position of the spot with the frame number, trigger number, and Z-axis position of the original point cloud output by the 3D profilometer 2. When the sampling time of the spot image and the sampling time of the point cloud are not completely consistent, the center position of the spot corresponding to the sampling time of the point cloud is obtained through nearest neighbor matching or interpolation, and then compared with the initial spot position to obtain the dynamic offset at the corresponding time, which is used for reverse subtraction compensation of the original point cloud. Thus, Figure 3 The light spot motion trajectory shown can be used to identify and quantify the lateral sway introduced by mechanical motion, and can also serve as a basis for calculating the dynamic offset at each sampling moment and restoring the true contour under the static reference coordinate system.

[0056] The high frame rate area array camera 5 continuously or triggeredly acquires no less than 1000 frames of spot images. The synchronization and compensation processing module 9 processes each frame of spot image: first, it extracts candidate spot regions based on brightness thresholds or band filtering results; then, it filters effective spot regions by connected component area, roundness, and grayscale peak value; finally, it calculates the sub-pixel center coordinates of the spot using the centroid method, Gaussian fitting, or elliptic fitting. .

[0057] Synchronization and compensation processing module according to and The difference is used to calculate the pixel displacement. , Based on the pre-calibrated proportional coefficient , or mapping matrix The pixel displacement is converted into the lateral offset of the 3D profilometer in actual space: , ; .

[0058] To reduce the impact of image noise and occasional misidentification, the synchronization and compensation processing module can perform median filtering, moving average filtering, Kalman filtering, or spline fitting on the coordinates of reference light points for 1000 or more consecutive frames to obtain the time-varying results. or Z-axis position Changing dynamic offset curve , or , .

[0059] When the 3D profilometer 2 outputs the raw point cloud, the synchronization and compensation processing module records the frame number, trigger number, and corresponding Z-axis position of each frame of point cloud, each scan line, or each point. The Z-axis position is preferably determined by the position count of a magnetic ruler, optical ruler, or encoder; if the sampling time of the spot image is not completely consistent with the sampling time of the point cloud, the dynamic offset at the corresponding time is calculated by nearest neighbor matching or interpolation.

[0060] For any point in the original point cloud If its sampling time is Then compensate the point for This yields the compensated point cloud. The compensated point cloud is equivalent to subtracting the lateral sway generated during mechanical movement from the point cloud coordinates in reverse, thereby restoring the true contour of the measured component 1 in the static reference coordinate system.

[0061] Figure 4 This is a schematic diagram comparing the point cloud and centerline before and after processing using the compensation method of this application. The horizontal axis in the figure represents the lateral displacement in the X direction (mm), and the vertical axis represents the position in the Z direction (mm).

[0062] Before compensation Figure 4 As shown on the left: During the mechanical movement, the 3D profilometer 2 outputs the original point cloud. The synchronization and compensation processing module simultaneously records the frame number, trigger number, and corresponding Z-axis position for each frame of point cloud, each scan line, or each point. This Z-axis position is preferably determined by the position count of a magnetic scale, optical scale, or encoder. When the sampling time of the light spot image is not completely consistent with the sampling time of the point cloud, the dynamic offset at the corresponding time is calculated through nearest neighbor matching or interpolation. Due to the lateral sway of the measured component 1 or the measurement system during continuous movement in the Z-direction, the original point cloud is superimposed with an offset component that varies with the Z-direction position in the X-direction. This is reflected in the centerline no longer remaining vertical, but instead exhibiting... Figure 4 The S-shaped (snake-like) curve shown in the left figure shows that the lateral dispersion of the point cloud relative to the true center line becomes more obvious as it approaches different heights.

[0063] For any point in the original point cloud, if its sampling time is t, the dynamic offset in the X direction at that time can be obtained by looking up the frame number, trigger number, and Z-axis position recorded during synchronization, or by interpolation. Then the compensated point can be taken as the point that is subtracted in the opposite direction in the X direction. (The Z-axis remains unchanged corresponding to the sampling position), thus completing the point-by-point correction from the original point to the compensated point. After the above compensation, the following is obtained: Figure 4 The point cloud after compensation shown on the right: the center line, which was originally distributed in an S-shape along the height direction, is straightened into a straight line that is approximately parallel to the Z-axis. After compensation, the center line runs through the center of the point cloud distribution, and the point cloud is basically symmetrically distributed around the center line.

[0064] Therefore, Figure 4 The "compensation" process from left to right essentially involves subtracting the lateral sway generated during mechanical motion from the point cloud coordinates in reverse, restoring the measured coordinates to the static reference coordinate system. The S-shaped bend in the left figure corresponds to the apparent deformation before correction of the X-direction dynamic offset caused by the motion. The vertical center line and the converged point cloud in the right figure demonstrate that the method of this application effectively eliminates the influence of lateral sway at different heights in the Z-direction motion on the contour measurement, restoring the true contour of the measured component 1 in the static reference coordinate system.

[0065] Example 2 Based on Embodiment 1, the height-measuring component 1 of the present invention can also be a slender component, or other parts with a height-direction extension structure that are sensitive to lateral accuracy.

[0066] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A dynamic error compensation method for three-dimensional detection of high-performance components based on a beam reference, characterized in that... include: S1. Establish the Z-axis position reference and beam reference, and calibrate the mapping relationship between the center pixel displacement of the spot image and the actual offset of the detection slide and the three-dimensional profilometer in the X and Y directions. S2, obtain the Z-axis position, and synchronously trigger the 3D profilometer and high frame rate area array camera according to the preset displacement, and establish a correspondence between the original 3D point cloud and the spot image according to at least one of the sampling time, frame number or Z-axis position. S3 drives the detection slide to scan the component being measured along the Z-axis, simultaneously acquiring the original three-dimensional point cloud and spot images; S4, extract the center coordinates of the light spot to generate the motion trajectory, and calculate the dynamic offset of the detection slide and the three-dimensional profilometer in the X and Y directions; S5, Match the dynamic offset to the original 3D point cloud according to the corresponding relationship; S6 performs reverse coordinate compensation on the original 3D point cloud to obtain the compensated 3D point cloud.

2. The dynamic error compensation method for three-dimensional detection of high-structure components based on a beam reference according to claim 1, characterized in that: The mapping relationship described in S1 is a proportional coefficient relationship, an affine mapping relationship, or a homography mapping relationship.

3. The dynamic error compensation method for three-dimensional detection of high-structure components based on a beam reference according to claim 1, characterized in that: In S1, the beam reference module is a laser collimator, laser collimator, or fixed beam generator that can emit a vertical beam or a collimated beam.

4. The dynamic error compensation method for three-dimensional detection of high-structure components based on a beam reference according to claim 1, characterized in that: In S4, the center coordinates of the light spots in 1000 or more consecutive light spot images are located at the sub-pixel level, and the dynamic offset curve is obtained by moving average filtering, median filtering, Kalman filtering or spline fitting.

5. The dynamic error compensation method for three-dimensional detection of high-structure components based on a beam reference according to claim 1, characterized in that: The beam reference module provides at least two reference light points, or the high frame rate area array camera has at least two; the translational deviation and angular deviation are calculated based on the imaging results of multiple reference light points or multiple high frame rate area array cameras, and a rigid body inverse coordinate transformation including translation and rotation is performed.

6. The dynamic error compensation method for three-dimensional detection of high-structure components based on a beam reference according to claim 1, characterized in that: Acquire a light spot image in a static state and determine the initial reference position; compare the real-time light spot center coordinates with the initial reference position to obtain the dynamic offset of the detection head and the three-dimensional profilometer in the X and Y directions.

7. The dynamic error compensation method for three-dimensional detection of high-structure components based on a beam reference according to claim 1, characterized in that: In S6, when only the translational deviations in the X and Y directions are calculated, two-dimensional lateral translation reverse compensation is performed on the original 3D point cloud; when the translational deviations and angular deviations are calculated, a rigid body reverse coordinate transformation including translations in the X, Y, and Z directions and rotations around at least one coordinate axis is performed on the original 3D point cloud.

8. A dynamic error compensation system for three-dimensional inspection of high-strength components based on a beam reference, characterized in that... include: A 3D profilometer is used to scan the height of a component being measured and output the original 3D point cloud. The detection slide moves the 3D profilometer up and down along the Z-axis. The beam reference module, independent of the detection slide and fixed to the stationary base, is used to emit a stable reference beam; A high frame rate area array camera is mounted on the same detection slide as the 3D profilometer and moves with the detection slide. The lens is oriented toward the reference beam so that the reference beam is incident and imaged to form a light spot. The synchronization and compensation processing module is used to match the original 3D point cloud with the spot image according to at least one of the sampling time, frame number or Z-axis position. It calculates the dynamic offset of the detection slide and 3D profilometer in the X and Y directions by the change of spot position, and compensates the original 3D point cloud coordinates in reverse to obtain the compensated 3D point cloud.

9. The dynamic error compensation system for three-dimensional detection of high-structure components based on a beam reference according to claim 8, characterized in that, It also includes a Z-axis position reference module and a hardware equidistant trigger module; the Z-axis position reference module is a magnetic grating ruler, optical grating ruler, encoder or linear displacement measuring device, and the hardware equidistant trigger module outputs a trigger signal according to position counting, so that the Z coordinate of the original three-dimensional point cloud is determined by the Z-axis position reference module.

10. The dynamic error compensation system for three-dimensional detection of high-structure components based on a beam reference according to claim 8, characterized in that, The beam reference module and the mechanical motion chain of the detection slide are independent of each other and are fixed to the foundation, independent reference base or external stabilizing support; the three-dimensional profilometer and the high frame rate area array camera are rigidly mounted on the same detection slide; the lens of the high frame rate area array camera is oriented towards the beam reference module, so that the stable reference beam is directly incident on the high frame rate area array camera and forms a light spot on the imaging surface.