Methods and Systems for Measuring the Geometric Elements of Automobile Wheel Hubs

Abstract

This invention discloses a method and system for measuring the geometric elements of an automobile wheel hub, comprising: projecting a coded pattern onto the wheel hub to be measured using a projector, and acquiring wheel hub images using an industrial camera to generate an initial point cloud of the wheel hub; denoising the initial point cloud of the wheel hub; clustering the denoised point cloud to extract the surface point cloud of the wheel hub; performing planar fitting on the surface point cloud of the wheel hub; compressing the surface point cloud of the wheel hub onto the fitted plane to obtain a planar point cloud, and then extracting the contour of the planar point cloud, extracting the surface contour point cloud from the surface point cloud of the wheel hub based on the index of the contour points; clustering the surface contour point cloud, with the largest cluster being the outer contour, the second largest cluster being the cover stop contour, and the remaining clusters being the bolt hole contours; scaling and projecting the cover stop contour and bolt hole contour to obtain the cover stop point cloud and bolt hole point cloud; and calculating the cover stop parameters and bolt hole parameters, as well as the positional degree between the bolt hole and the cover stop, based on the geometric features of the cover stop and bolt hole and the bolt hole point cloud.

Description

Technical Field

[0001] This invention relates to a method and system for measuring the geometric elements of automobile wheel hubs, belonging to the field of wheel hub inspection technology. Background Technology

[0002] The wheel hub is the connecting part between the car wheel and the body, playing a crucial role in supporting the overall weight of the vehicle, cushioning external impacts, and bearing lateral loads, driving forces, and braking torques. The center hole and bolt holes of the wheel hub are key geometric elements for positioning and connection; their dimensional and positional accuracy is of paramount importance to safe vehicle operation. Deformation of the wheel hub can lead to vehicle instability, accelerated tire wear, and increased road noise. Therefore, accurately and efficiently measuring the key geometric elements of the wheel hub is an essential prerequisite for ensuring wheel hub performance and maintaining vehicle driving safety.

[0003] Currently, common wheel hub measurement and inspection solutions calculate wheel hub parameters using two-dimensional vision. First, an industrial camera is used to acquire wheel hub images, and then traditional image processing algorithms are used to calculate the wheel hub's geometric parameters. For example, the "Aluminum Alloy Wheel Hub Dimension Measurement Based on Machine Vision" proposed by Du Lifeng, Zhou Zheng, and others uses the two-dimensional Canny edge detection algorithm to obtain the wheel hub contour, and then uses the Hough algorithm to fit the contour and measure the outer diameter. The "Research on Automatic Measurement System of Automobile Wheel Hub Dimensions Based on Machine Vision" proposed by Liu Tao, Luo Yubin, and others obtains wheel hub dimension data through image processing, feature extraction, and data fitting. The "An Intelligent Method for Rapid Measurement of Wheel Hub Model Dimensions" proposed by Xie Hui, Geng Jiaqin, and others processes the top and cross-sectional views of the wheel hub to extract the wheel hub contour and determine the coordinates, thereby calculating the wheel hub's diameter, width, and other dimensions based on the coordinates. Although two-dimensional vision has been developed earlier, is technologically mature, and has lower costs, it is easily affected by lighting interference and the texture of the wheel hub itself, resulting in larger measurement errors.

[0004] Two-dimensional vision often has limitations when dealing with complex wheel hub structures and high-precision requirements. With continuous technological advancements, three-dimensional vision methods have brought revolutionary changes to wheel hub inspection. It breaks through the constraints of traditional measurement, capturing the three-dimensional morphology of the wheel hub in a non-contact, high-precision, and high-speed manner. This provides a reliable data foundation for optimizing wheel hub manufacturing processes, quality assessment, and new product development, leading the field of wheel hub measurement to new heights. In recent years, with the development and popularization of LiDAR and 3D scanning technologies, these technologies have been widely applied in various fields. Benefiting from these technological advancements, three-dimensional vision-based measurement technology has emerged. This technology primarily employs non-contact measurement, avoiding deformation and damage to the measured surface, and can obtain complete three-dimensional data of the object being inspected. Currently, there are some studies on measuring wheel hub geometric elements using three-dimensional vision, but most of them convert the three-dimensional data acquired by the scanner into two-dimensional images, resulting in relatively few measured geometric elements. Summary of the Invention

[0005] To quickly and accurately measure the geometric parameters of wheel hubs, this invention proposes a measurement method based on three-dimensional vision. This method directly processes three-dimensional point clouds and can quickly and accurately measure eight geometric elements, including the diameter of the wheel hub cover stop, the diameter of the bolt holes, and the positional accuracy. This solves the problem that existing three-dimensional vision measurement methods can only measure a limited number of wheel hub geometric elements.

[0006] To address the above problems, one aspect of the present invention proposes the following technical solution:

[0007] A method for measuring the geometric elements of an automobile wheel hub includes the following steps: S1, projecting a coded pattern onto the wheel hub to be measured and acquiring a wheel hub image to generate an initial point cloud of the wheel hub; S2, denoising the initial point cloud of the wheel hub to obtain a denoised point cloud; S3, performing clustering processing on the denoised point cloud to extract the surface point cloud of the wheel hub; S4, performing planar fitting on the surface point cloud of the wheel hub to obtain a fitting plane; S5, compressing the surface point cloud of the wheel hub onto the fitting plane to obtain a planar point cloud, and then performing contour extraction on the planar point cloud, extracting the surface contour point cloud from the surface point cloud of the wheel hub according to the index of the contour points; S6, using clustering processing to divide the surface contour point cloud into clusters, among which the most... The largest cluster represents the outer contour, the second largest cluster represents the cap stop contour, and the remaining clusters represent bolt hole contours; S7, the cap stop contour and the bolt hole contour are scaled and projected to obtain cap stop point clouds and bolt hole point clouds respectively; S8, based on the geometric features of the cap stop, cap stop parameters are calculated based on the cap stop point clouds, including cap stop diameter, center hole diameter, cap groove depth, and wheel core thickness; based on the geometric features of the bolt holes, bolt hole parameters are calculated based on the bolt hole point clouds, including bolt hole countersunk diameter, bolt hole diameter, and bolt hole depth; the distance between the bolt hole contour and the cap stop contour is calculated to obtain the positional accuracy of the bolt hole and the cap stop.

[0008] In another aspect, this invention proposes a measurement system for the geometric elements of an automotive wheel hub, comprising: a projector for projecting a coded pattern onto the wheel hub to be measured; an industrial camera for acquiring an image of the wheel hub to be measured with the coded pattern projected onto it; and a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, performs the following steps: reconstructing a point cloud based on the image of the wheel hub to be measured acquired by the industrial camera to obtain an initial point cloud of the wheel hub; denoising the initial point cloud of the wheel hub to obtain a denoised point cloud; performing clustering processing on the denoised point cloud to extract point clouds on the surface of the wheel hub; performing plane fitting on the point cloud of the surface of the wheel hub to obtain a fitted plane; compressing the point cloud of the surface of the wheel hub onto the fitted plane to obtain a planar point cloud, and then extracting the contour of the planar point cloud, based on the contour points... The surface contour point cloud is extracted from the wheel hub surface point cloud; the surface contour point cloud is divided into clusters using clustering processing, where the largest cluster is the outer contour, the second largest cluster is the cap stop contour, and the remaining clusters are bolt hole contours; the cap stop contour and the bolt hole contour are scaled and projected to obtain cap stop point clouds and bolt hole point clouds respectively; based on the geometric features of the cap stop, cap stop parameters are calculated based on the cap stop point clouds, including cap stop diameter, center hole diameter, cap groove depth, and wheel core thickness; based on the geometric features of the bolt holes, bolt hole parameters are calculated based on the bolt hole point clouds, including bolt hole countersunk diameter, bolt hole diameter, and bolt hole depth; the distance between the bolt hole contour and the cap stop contour is calculated to obtain the positional accuracy of the bolt hole and the cap stop.

[0009] Compared with existing technologies, the beneficial effects of this invention are mainly reflected in the following aspects: This invention acquires three-dimensional point cloud data of the wheel hub through three-dimensional vision, and processes the three-dimensional point cloud data through steps S2 to S8 to calculate multiple cap stop parameters, multiple bolt hole parameters, and the positional accuracy of bolt holes and cap stops from the three-dimensional point cloud data. Compared with the existing technology that converts three-dimensional point clouds into two-dimensional images and then measures geometric elements through image processing, this invention avoids the complex image processing process and the impact of errors during the conversion to two-dimensional images on the accuracy of subsequent geometric element measurements. Furthermore, it can measure more wheel hub geometric elements. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the automotive wheel hub geometry measurement system according to an embodiment of the present invention;

[0011] Figure 2 This is a flowchart of the method for measuring the geometric elements of automobile wheel hubs according to an embodiment of the present invention;

[0012] Figure 3 This is a schematic diagram illustrating the calculation process of automotive wheel hub geometric elements according to an embodiment of the present invention;

[0013] Figure 4 This is a cross-sectional view of a car wheel hub;

[0014] Figure 5 This is a schematic diagram of the depth of the groove in the car wheel hub cap. Detailed Implementation

[0015] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are provided for illustrative purposes only and are not intended to limit the scope of protection of the present invention in any way.

[0016] A specific embodiment of the present invention provides a method for measuring the geometric elements of an automobile wheel hub. This method first establishes a... Figure 1 The measurement system shown includes a DLP projector 10, an industrial camera 20, and a computer 30. Based on this system, the DLP projector 10 projects a preset coded pattern onto the wheel hub 40 to be measured, the industrial camera 20 acquires an image of the wheel hub, and the computer 30 generates initial point cloud data of the wheel hub. After a series of processing steps, multiple geometric elements of the wheel hub are calculated from this data. Based on this understanding, the computer 30 should include a computer-readable storage medium storing a predetermined computer program. When executed by a processor, this predetermined computer program should be able to perform the following steps:

[0017] Phase calculation is performed on the wheel hub image acquired by the industrial camera, and the calibration parameters obtained after calibration are loaded to complete the point cloud reconstruction and generate the initial point cloud of the wheel hub.

[0018] The initial point cloud of the wheel hub is denoised to obtain a denoised point cloud;

[0019] Clustering is performed on the denoised point cloud to extract the point cloud from the wheel hub surface;

[0020] Perform plane fitting on the point cloud of the wheel hub surface to obtain the fitting plane;

[0021] The wheel hub surface point cloud is compressed onto the fitting plane to obtain a planar point cloud. Then, the planar point cloud is contour extracted, and the surface contour point cloud is extracted from the wheel hub surface point cloud according to the index of the contour points.

[0022] Clustering is used to divide the surface contour point cloud into clusters, where the largest cluster is the outer contour, the second largest cluster is the cover stop contour, and the remaining clusters are bolt hole contours.

[0023] The cover stop contour and the bolt hole contour are scaled and projected to obtain the cover stop point cloud and the bolt hole point cloud, respectively.

[0024] Based on the geometric features of the cap stop, the cap stop parameters are calculated using the cap stop point cloud, including the cap stop diameter, center hole diameter, cap groove depth, and wheel core thickness. Based on the geometric features of the bolt holes, the bolt hole parameters are calculated using the bolt hole point cloud, including the bolt hole countersunk diameter, bolt hole diameter, and bolt hole depth. The distance between the bolt hole profile and the cap stop profile is calculated to obtain the positional accuracy between the bolt hole and the cap stop.

[0025] Based on the above system, a specific embodiment of the present invention also proposes a method for measuring the geometric elements of automobile wheel hubs, as described above. Figures 1 to 3 The measurement method includes the following steps S1 to S8:

[0026] S1. Obtaining the Initial Point Cloud: First, a white non-reflective backplate 50 is installed on the inner side of the wheel hub 40 under test, ensuring that the center hole and five bolt holes of the hub are all within the backplate area, and the backplate plane is tightly fitted to the inner side of the wheel hub. Then, a predetermined coded pattern (such as stripes) is projected onto the wheel hub 40 under test using a DLP projector 10. Next, an image of the hub is acquired using a calibrated industrial camera 20. Finally, phase calculation is performed on the hub image acquired by the industrial camera, and the calibration parameters obtained after calibration are loaded to complete point cloud reconstruction, generating the initial point cloud of the hub. Existing technologies can be used for point cloud reconstruction based on the images acquired by the camera, which will not be elaborated here. Furthermore, the plane of the projector lens is parallel to the backplate.

[0027] In a preferred embodiment of the invention, for bolt holes that are deep, a backing plate and multiple exposure methods are used. A DLP projector and an industrial camera are used to perform three-dimensional measurement of the bolt holes. The principle is to project the bolt holes and acquire multiple sets of stripe image sequences at different exposure times, and then fuse them into a single image for the calculation of the three-dimensional point cloud. Furthermore, to address the problem that the height information of the inner wall of the measured object cannot be well obtained due to the depth of the bolt holes and the severe reflection of the inner wall, we indirectly obtain the height information of the bolt holes by reconstructing the plane of the white backing plate, which facilitates the subsequent calculation of various geometric elements.

[0028] S2. Point cloud preprocessing: Denoise the initial point cloud of the wheel hub obtained in step S1 to obtain a denoised point cloud.

[0029] Noise, also known as outliers, refers to irrelevant or unwanted interference signals or errors in point cloud data. It is usually caused by factors such as the brightness of ambient light, the accuracy of measurement equipment and system errors, the texture of object materials and surfaces, and human jitter. Noise needs to be preprocessed to remove it, thereby improving the calculation accuracy.

[0030] Statistical filtering can effectively remove isolated noise points that do not conform to the overall point cloud distribution, thus making the point cloud data smoother and more accurate. Therefore, in some embodiments of the present invention, statistical filtering is used to denoise the initial point cloud of the wheel hub to obtain a denoised point cloud, sor_filtered_cloud. It should be understood that the present invention can also use other point cloud denoising methods, and the present invention is not limited thereto.

[0031] Statistical filtering principle:

[0032] The distances between each point and its n neighboring points form a Gaussian distribution, the shape of which is determined by the mean u and the standard deviation σ. Let the coordinates of the i-th point in the point cloud be p. i (x i ,y i , z i ), the distance from this point to any neighboring point p n (x n ,y n , z n The distance is:

[0033]

[0034] The formula for calculating the average distance from each point to any neighboring point is:

[0035]

[0036] The standard deviation is:

[0037]

[0038] Let std be the standard deviation factor. Based on the set thresholds k and std, a point is retained when the average distance of its k nearest neighbors is within the standard range (μ-σ·std, μ+σ·std). Points outside this range are defined as outliers and deleted.

[0039] S3. Extract the wheel hub surface point cloud: Use Euclidean clustering to cluster the denoised point cloud sor_filtered_cloud. This process groups similar contours into clusters based on their spatial approximation, with the largest cluster being the wheel hub surface point cloud surface_cloud.

[0040] S4. Adjusting Point Cloud Pose: The Random Sample Consensus (RANSAC) algorithm is used to perform planar fitting on the wheel hub surface point cloud (surface_cloud) to obtain the fitting plane P_surface. Specifically, firstly, the angle between the normal vector of the wheel hub surface point cloud (surface_cloud) and the Z-axis of the world coordinate system is calculated to obtain the rotation matrix; then, this rotation matrix is ​​used to rotate the wheel hub surface point cloud (surface_cloud) and the denoised point cloud (sor_filtered_cloud) to adjust the pose of the surface_cloud and sor_filtered_cloud to a state parallel to the XOY plane of the world coordinate system, aiming to reduce the complexity of subsequent processing.

[0041] S5. Contour Extraction: Compress the wheel hub surface point cloud surface_cloud onto the fitting plane P_surface to obtain the planar point cloud plane_surface_cloud. Then, use the alpha-shape algorithm to extract the contour of the planar point cloud plane_surface_cloud. Based on the index of the contour points, extract the surface contour point cloud surface_cloud_hull from the wheel hub surface point cloud surface_cloud.

[0042] S6. Automatically identify and label the point cloud of the cover stop and bolt holes: Euclidean clustering is used to divide the surface contour point cloud surface_cloud_hull into clusters, where the largest cluster is the outer contour, the second largest cluster is the cover stop contour center_circle, and the remaining clusters are the bolt hole contours bolt_circle. Since there are multiple bolt holes, in this step, the bolt hole contours bolt_circle can be sorted clockwise according to their centroid positions.

[0043] S7. Precise Segmentation of the Cover Edge and Bolt Hole Point Cloud: To avoid interference from the inner wall point cloud on the calculation of wheel hub geometry, this embodiment of the invention designs a scaling projection segmentation algorithm to accurately segment the cover edge and bolt hole point cloud. Details are as follows:

[0044] S71. Shrink and adjust the cover stop contour center_circle to obtain a new cover stop contour new_center_circle. Then extract the corresponding point cloud of the world coordinate system XOY plane projection points in the denoised point cloud sor_filtered_cloud in new_center_circle to obtain the cover stop point cloud center_cloud.

[0045] S72. Shrink and adjust the bolt hole outline bolt_circle to obtain a new bolt hole outline new_bolt_circle. Then extract the corresponding point cloud of the world coordinate system XOY plane projection points in the denoised point cloud sor_filtered_cloud in new_bolt_circle to obtain the bolt hole point cloud bolt_cloud.

[0046] The principle behind the "shrinkage adjustment" described in S71 and S72 above is the same, as follows:

[0047] 1) Calculate the centroid (center of the circle): The centroid is the average position of all points on the circle, which is obtained by calculating the average position of all points on each coordinate axis;

[0048] 2) Determine the scaling factor: Select a scaling factor based on the radius of the target circle to reduce the size of the circle;

[0049] 3) Create a new point cloud: Prepare a new set to store the points of the scaled-down circle;

[0050] 4) Shrinking the point cloud of a circle: Traverse all points on the circle and shrink each one individually:

[0051] 4.1) For each point, first calculate the position offset of that point relative to the centroid, that is, the distance of that point from the centroid;

[0052] 4.2) Multiply this distance by a scaling factor to reduce the distance;

[0053] 4.3) Based on the reduced distance, recalculate the new position of the point, which is the centroid plus the reduced distance vector.

[0054] 4.4) Add the newly calculated location points to the new point set to obtain the point cloud of the reduced circle.

[0055] The "circle" mentioned above refers to the center_circle of the cover stop and the bolt_circle of several bolt holes. The center_circle and the bolt_circle are reduced in size through the above steps.

[0056] Finally, extract the XOY projection points of the target point cloud (i.e., the denoised point cloud sor_filtered_cloud) within the scaled-down circle.

[0057] S8. Calculate various parameters of the wheel hub based on its geometric characteristics: such as Figure 4 and Figure 5In this embodiment of the invention, the geometric parameters of the wheel hub to be measured include the diameter of the cap stop 1, the diameter of the center hole 2, the depth of the cap groove 3, the thickness of the wheel core 4, the diameter of the bolt hole countersunk hole 5, the diameter of the bolt hole 6, the depth of the bolt hole 7, and the positional degree between the bolt hole and the cap stop.

[0058] S81. Calculation of cap stop parameters, including calculation of cap stop diameter 1, center hole diameter 2, cap groove depth 3, and wheel core thickness 4, as follows:

[0059] S811. Euclidean clustering is used to cluster the point cloud center_cloud of the cover stop, which is divided into two categories: the largest cluster and the remaining clusters. The largest cluster is the point cloud center_2 of the cover stop back plate, and the remaining clusters are the point cloud center_1 of the cover stop countersunk hole surface.

[0060] S812. By performing a circle fitting operation on the cover stop profile center_circle using the least squares method, the diameter of the cover stop can be obtained.

[0061] S813. Use the alpha-shape algorithm to extract the center hole contour from the point cloud center_2 of the cover stop back plate and perform a circle fitting operation to obtain the center hole diameter.

[0062] S814. Using the RANSAC algorithm, perform plane fitting on the cap stop contour center_circle and the point cloud center_1 of the cap stop countersunk hole surface, and calculate the distance between the two planes to obtain the cap groove depth.

[0063] S815. Perform plane fitting on the cover stop contour center_circle and the cover stop back plate point cloud center_2 respectively, and calculate the distance between the two planes to obtain the wheel core thickness.

[0064] S82. Bolt hole parameter calculation, including calculating the countersunk diameter, bolt hole diameter, and bolt hole depth; as follows:

[0065] S821. Euclidean clustering is used to cluster the bolt hole point cloud bolt_cloud into two categories: the largest cluster and the remaining clusters. The largest cluster is the bolt hole back plate point cloud bolt_2, and the remaining clusters are the bolt hole countersunk surface point cloud bolt_1.

[0066] S822. Perform a circle fitting operation on the bolt hole outline bolt_circle based on the least squares method to obtain the countersunk diameter of the bolt hole;

[0067] S823. Use the alpha-shape algorithm to extract the contour from the bolt hole back plate point cloud bolt_2 and perform circle fitting to obtain the bolt hole diameter;

[0068] S824. Using the RANSAC algorithm, perform plane fitting on the bolt hole back plate point bolt_2 and the bolt hole countersunk surface point cloud bolt_1 respectively, and calculate the distance between the two planes to obtain the bolt hole depth.

[0069] S83. Calculate the positional accuracy between the bolt hole and the cap stop:

[0070] Calculate the distance from the centroid of each bolt_circle to the centroid of center_circle to obtain the positional accuracy between the bolt hole and the cap stop. For example... Figure 3 As shown, there are five bolt circles corresponding to bolt holes, namely bolt circle 1, bolt circle 2, bolt circle 3, bolt circle 4, and bolt circle 5.

[0071] In this embodiment of the invention, the projector and industrial camera are calibrated before use. The calibration parameters of phase-three-dimensional coordinates are obtained using the polynomial model method for subsequent reconstruction. After calibration, the coded stripe image is projected onto the wheel hub to be tested using the projector. Then, the industrial camera acquires images and obtains the absolute phase using the phase shift method and the multi-frequency heterodyne method. Finally, the absolute phase is substituted into the previously obtained calibration parameters, and a high-quality initial point cloud of the wheel hub, input_cloud, is calculated from the phase information.

[0072] The steps for obtaining the calibration parameters of the phase-3D coordinates using the polynomial model method include:

[0073] Step 1: Place the calibration plate at any position within the measurement range, photograph the calibration plate with an industrial camera, detect the corner points of the calibration plate, and record the coordinates (u) of each corner point in the imaging plane. ci v ci (i = 1, 2, 3, ..., n);

[0074] Step two: Keeping the calibration plate stationary, project the sine fringe pattern onto the calibration plate using a projector to obtain the phase diagram at this point, and record the phase Φ at each corner point. i (i = 1, 2, 3, ..., n);

[0075] Step 3: Adjust the placement of the calibration plate multiple times, repeating Step 1 and Step 2 15 times to obtain multiple sets of corner coordinates (u). ci v ci ) and the corresponding phase Φ i ;

[0076] Step 4: Calibrate the industrial camera using Zhang Zhengyou's camera calibration method to obtain the intrinsic parameters of the industrial camera and the extrinsic parameters corresponding to each calibration board image. Based on the extrinsic parameters, the coordinates of the corner points in each calibration board image in the industrial camera coordinate system can be calculated.

[0077] Step 5: Using the phase of the calibration board corner point position obtained in Step 3 and the phase of the calibration board corner point (u) obtained in Step 4... ci v ci The coordinates (X) in the industrial camera coordinate system ci Y ci Z ci Constructing polynomials

[0078] a1X c +a2Y c +a3Z c +a4-a5ΦX c -a6ΦY c -a7ΦZ c -a8Φ=0

[0079] The eight parameters a1-a8 were obtained using the least squares method.

[0080] Phase shift method principle:

[0081] In the standard N-step phase-shifting method, the grayscale value of the i-th sinusoidal grating image acquired by the camera can be expressed as:

[0082] I i (x,y)=A(x,y)+B(x,y)cos[φ(x,y)+α i ], i = 1, 2, 3, ..., N (3-1)

[0083] Among them, I i (x,y) represents the grayscale value at point (x,y) in the image captured by the camera during the i-th phase shift; A(x,y) is the background light intensity value at this point; B(x,y) represents its modulation intensity value; φ(x,y) is the principal phase value at this point, which is related to the spatial frequency f0 of the sinusoidal fringe and the current coordinates (x,y); α i The phase shift increment of the i-th phase shift map can be represented by α. i =2π(i-1) / N represents I. For a single pixel, I i Since A, B, and φ are known, while A, B, and φ are unknown, the phase principal value φ can be calculated using the least squares algorithm when N≥3:

[0084]

[0085] Substituting equation (3-1) into equation (3-2), we get:

[0086]

[0087] because and Therefore, the representation of the principal phase value can be simplified as follows:

[0088]

[0089] Principle of multi-frequency heterodyne method:

[0090] In the multi-frequency heterodyne method, the phase principal values ​​of multiple sets of fringe patterns with different spatial frequencies are subtracted to obtain the phase principal value of a lower spatial frequency distribution. This phase difference is designed to cover the entire measurement range within one period. Finally, the absolute phase information is obtained based on the phase difference and its spatial frequency multiple relationship with the original phase principal value. For example, in the 3-frequency heterodyne method, three sets of fringes with different spatial frequencies are projected, with spatial periods T1, T2, and T3, and phase principal values ​​φ1, φ2, and φ3, respectively. The unfolded absolute phases are as follows: In practical applications, only the absolute phase corresponding to a certain frequency is needed; here, we take it as the final target. In obtaining First, it is necessary to heterodyne the principal phase values ​​of the fringes with spatial periods T1 and T2 and the principal phase values ​​of the fringes with periods T2 and T3 to obtain φ. 12 φ 23 Then, perform the same operation on φ 12 φ 23 φ is obtained by heterodyne. 123 According to the principle of heterodyne calculation, φ 12 φ 23 φ 123 Their respective expressions are

[0091]

[0092]

[0093]

[0094] The corresponding spatial periods are

[0095]

[0096] absolute phase It needs to be derived from the corresponding principal phase value φ1 and fringe order k1:

[0097]

[0098]

[0099] The Round function rounds down integers to the nearest integer, and in it... It is necessary to obtain the corresponding phase principal value φ 12 and stripe level k 12 The derivation yields:

[0100]

[0101]

[0102] Finally, by substituting the previously obtained calibration parameters, high-quality 3D point cloud data is calculated from the phase information:

[0103] Given any point (u) c v c ) and the phase Φ corresponding to that point, as well as the camera's intrinsic parameters (f cx f cy c cx c cy The final three-dimensional coordinates can be obtained using the following formula:

[0104]

[0105] Euclidean clustering principle:

[0106] 1. Select any point p from the input point cloud. i (x i ,y i , z i ), calculate the neighborhood points within a preset radius r;

[0107] 2. Calculate the value of each neighboring point p. n (x n ,y n , z n Distance to that point:

[0108]

[0109] 3. Nodes within a neighborhood whose distance is less than a set threshold are clustered into set Q;

[0110] 4. If the number of elements in Q no longer increases, the entire clustering process ends; otherwise, p must be selected from set Q. i For points other than Q, repeat the above process until the number of elements in Q stops increasing.

[0111] 5. Repeat steps 1-4 until all points are clustered.

[0112] RANSAC (Random Sample Consensus) principle:

[0113] 1. As we know from mathematical knowledge, at least three points are needed to fit a plane. Therefore, we first randomly select three points from the input point cloud (the point cloud that needs to be fitted to the plane using the RANSAC algorithm), and then calculate the plane model parameters A, B, C, and D according to the following plane equation Ax + By + Cz + D = 0.

[0114] 2. Use the remaining data points to test the estimated plane model in the plane equation, calculate the error, compare the error with the set error threshold, and if it is less than the set threshold, then determine the point as an interior point, count the number of interior points under this parameter model and record it.

[0115] 3. Continue with steps 1 and 2. If the number of inliers in the current model is greater than the maximum number of inliers that have been saved, update the model parameters. The model parameters that are saved are always the model parameters with the most inliers.

[0116] 4. Repeat steps 1 to 3, iterating continuously until the iteration threshold is reached. Find the model parameter with the most interior points, and finally estimate the model parameter again using the interior points to obtain the final model parameter.

[0117] Principle of alpha-shape algorithm:

[0118] (1) From the input point set P (i.e., the point cloud that needs to be processed by the alpha-shape algorithm), any point p i (x i ,y i Starting from a point p, within a neighborhood set R of a preset radius of 2a, take any point p. n (x n ,y n ), calculate the result from point p i (x i ,y i ) and point p n (x n ,y n The center p of the circle is determined. o (x o ,y o );

[0119]

[0120]

[0121] (2) Calculate the distance from other points in the neighborhood point set R to the center p of the circle. o (x o ,y o The distances to p are: if all distances are greater than a, it means there are no other points inside the circle, then p... i pn p is the boundary line segment i p n These are boundary contour points; otherwise, they are non-boundary line segments.

[0122] (3) Repeat the first two steps for the remaining points in the neighborhood point set R until all points in R have been judged.

[0123] (4) Repeat the above steps for the remaining points in the point set P until all points in P have been judged.

[0124] The method for measuring the geometric elements of a car wheel hub in this invention removes noise through statistical filtering for bolt holes with high surface reflectivity and deep depth. Euclidean clustering is used to accurately extract and segment the point cloud of the wheel hub surface. The application of the RANSAC algorithm adjusts the point cloud to be perpendicular to the Z-axis of the world coordinate system, improving the accuracy of the analysis. Alpha-shape and scaling projection segmentation algorithms are used to accurately segment the point cloud of the wheel hub cover and each bolt hole, enhancing the calculation accuracy of key parameters.

[0125] The accuracy of key geometric elements detection in the wheel hub reaches 100 micrometers. The combination of high-precision point cloud data, accurate segmentation of the cover stop and bolt holes, and RANSAC algorithm fitting of planes and circles ensures that the accuracy of key geometric elements detection can reach 100 micrometers.

[0126] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, several equivalent substitutions or obvious modifications can be made without departing from the concept of the present invention, and all such modifications, achieving the same performance or purpose, should be considered within the scope of protection of the present invention.

Claims

1. A method for measuring the geometric elements of an automobile wheel hub, characterized in that, The following steps are involved: S1. Project the coded pattern onto the wheel hub to be tested and acquire the wheel hub image to generate the initial point cloud of the wheel hub; S2. Denoise the initial point cloud of the wheel hub to obtain a denoised point cloud; S3. Perform clustering processing on the denoised point cloud to extract the point cloud on the wheel hub surface; S4. Perform plane fitting on the point cloud of the wheel hub surface to obtain the fitting plane; S5. Compress the wheel hub surface point cloud onto the fitting plane to obtain a planar point cloud, then extract the contour of the planar point cloud, and extract the surface contour point cloud from the wheel hub surface point cloud according to the index of the contour points. S6. The surface contour point cloud is divided into clusters using clustering processing, where the largest cluster is the outer contour, the second largest cluster is the cover stop contour, and the remaining clusters are bolt hole contours. S7. The cover stop contour and the bolt hole contour are scaled and projected to obtain cover stop point cloud and bolt hole point cloud, respectively. The scaling and projection segmentation includes: scaling down the cover stop contour to obtain a new cover stop contour, and then extracting the corresponding point cloud of the world coordinate system XOY plane projection points in the new cover stop contour from the denoised point cloud to obtain the cover stop point cloud; scaling down the bolt hole contour to obtain a new bolt hole contour, and then extracting the corresponding point cloud of the world coordinate system XOY plane projection points in the new bolt hole contour from the denoised point cloud to obtain the bolt hole point cloud. The reduction adjustment of the cover stop profile specifically includes: multiplying the distance of each point on the cover stop profile relative to the centroid of the cover stop profile by a preset first scaling factor to obtain the reduced distance vector of each point on the cover stop profile; adding the centroid of the cover stop profile to the reduced distance vector of each point on the cover stop profile to obtain the new position of each point on the cover stop profile; the set of the new positions of each point on the cover stop profile is the new cover stop profile. The bolt hole contour is reduced in size by: multiplying the distance of each point on the bolt hole contour relative to the centroid of the bolt hole contour by a preset second scaling factor to obtain the reduced distance vector of each point on the bolt hole contour; adding the centroid of the bolt hole contour to the reduced distance vector of each point on the bolt hole contour to obtain the new position of each point on the bolt hole contour; the set of the new positions of each point on the bolt hole contour is the new bolt hole contour. S8. Based on the geometric features of the cap stop, calculate the cap stop parameters based on the cap stop point, the cap stop parameters including the cap stop diameter, the center hole diameter, the cap groove depth, and the wheel core thickness; based on the geometric features of the bolt hole, calculate the bolt hole parameters based on the bolt hole point, the bolt hole parameters including the bolt hole countersunk diameter, the bolt hole diameter, and the bolt hole depth; calculate the distance between the bolt hole profile and the cap stop profile to obtain the positional accuracy between the bolt hole and the cap stop.

2. The method for measuring the geometric elements of an automobile wheel hub as described in claim 1, characterized in that, Before step S1, the process also includes: S0, adding a white non-reflective backplate to the inner side of the wheel core of the wheel hub to be tested, with the backplate plane fitting against the inner side of the wheel core.

3. The method for measuring the geometric elements of an automobile wheel hub as described in claim 1, characterized in that, Step S3 includes: performing clustering processing on the denoised point cloud using Euclidean clustering, and grouping them into clusters according to spatial approximation, wherein the largest cluster is the point cloud of the wheel hub surface.

4. The method for measuring the geometric elements of an automobile wheel hub as described in claim 1, characterized in that, Step S4 involves performing planar fitting on the point cloud of the wheel hub surface, specifically including: The rotation matrix is ​​obtained by calculating the angle between the normal vector of the point cloud on the hub surface and the Z-axis of the world coordinate system using the random sample consensus algorithm. The rotation matrix is ​​used to rotate the point cloud on the wheel hub surface and the denoised point cloud to adjust their poses to be parallel to the world coordinate system XOY plane.

5. The method for measuring the geometric elements of an automobile wheel hub as described in claim 2, characterized in that, Step S8 involves calculating the cap stop parameters based on the geometric features of the cap stop and the cap stop points, specifically including: The point cloud of the cover stop is clustered into two categories: the largest cluster and the remaining clusters. The largest cluster is the point cloud of the cover stop back plate, and the remaining clusters are the point clouds of the cover stop countersunk hole surface. Perform a circle fitting operation on the contour of the cover stop to obtain the diameter of the cover stop; The outline of the center hole is extracted from the point cloud of the cover stop back plate and a circle fitting operation is performed to obtain the diameter of the center hole; The depth of the cap groove is obtained by performing plane fitting on the point cloud of the cap stop contour and the countersunk hole surface of the cap stop, and calculating the distance between the two planes. The thickness of the wheel core is obtained by performing planar fitting on the contour of the cover stop and the point cloud of the cover stop back plate, and calculating the distance between the two planes.

6. The method for measuring the geometric elements of an automobile wheel hub as described in claim 2, characterized in that, Step S8 involves calculating bolt hole parameters based on the bolt hole geometric features and the bolt hole point data, specifically including: The bolt hole point cloud is clustered into two categories: the largest cluster and the remaining clusters. The largest cluster is the bolt hole back plate point cloud, and the remaining clusters are the bolt hole countersunk surface point cloud. Perform a circle fitting operation on the bolt hole profile to obtain the countersunk diameter of the bolt hole; The bolt hole back plate point cloud is extracted for contour, and then a circle fitting operation is performed on the contour to obtain the bolt hole diameter; The bolt hole depth is obtained by performing plane fitting on the point cloud of the bolt hole back plate and the point cloud of the bolt hole countersunk surface, and calculating the distance between the two planes.

7. The method for measuring the geometric elements of an automobile wheel hub as described in claim 2, characterized in that, Step S8 calculates the distance between the bolt hole profile and the cover stop profile to obtain the positional accuracy of the bolt hole and the cover stop, specifically including: For each bolt hole of the hub under test, calculate the distance from the centroid of the bolt hole profile to the centroid of the cover stop profile to obtain the positional accuracy of each bolt hole relative to the cover stop.

8. A system for measuring the geometric elements of an automobile wheel hub, characterized in that, include: A projector is used to project coded patterns onto the wheel hub to be tested. An industrial camera is used to capture images of a wheel hub to be tested, which is projected with a coded pattern. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps: Point cloud reconstruction is performed on the images of the wheel hub to be tested acquired by an industrial camera to obtain the initial point cloud of the wheel hub; The initial point cloud of the wheel hub is denoised to obtain a denoised point cloud; Clustering is performed on the denoised point cloud to extract the point cloud from the wheel hub surface; Perform plane fitting on the point cloud of the wheel hub surface to obtain the fitting plane; The wheel hub surface point cloud is compressed onto the fitting plane to obtain a planar point cloud. Then, the planar point cloud is contour extracted, and the surface contour point cloud is extracted from the wheel hub surface point cloud according to the index of the contour points. Clustering is used to divide the surface contour point cloud into clusters, where the largest cluster is the outer contour, the second largest cluster is the cover stop contour, and the remaining clusters are bolt hole contours. The cover stop contour and the bolt hole contour are respectively scaled and projected to obtain cover stop point cloud and bolt hole point cloud; the scaling and projection segmentation includes: scaling down and adjusting the cover stop contour to obtain a new cover stop contour, and then extracting the corresponding point cloud of the world coordinate system XOY plane projection points in the new cover stop contour from the denoised point cloud to obtain the cover stop point cloud; the bolt hole contour is scaled down and adjusted to obtain a new bolt hole contour, and then extracting the corresponding point cloud of the world coordinate system XOY plane projection points in the new bolt hole contour from the denoised point cloud to obtain the bolt hole point cloud. The reduction adjustment of the cover stop profile specifically includes: multiplying the distance of each point on the cover stop profile relative to the centroid of the cover stop profile by a preset first scaling factor to obtain the reduced distance vector of each point on the cover stop profile; adding the centroid of the cover stop profile to the reduced distance vector of each point on the cover stop profile to obtain the new position of each point on the cover stop profile; the set of the new positions of each point on the cover stop profile is the new cover stop profile. The bolt hole contour is reduced in size by: multiplying the distance of each point on the bolt hole contour relative to the centroid of the bolt hole contour by a preset second scaling factor to obtain the reduced distance vector of each point on the bolt hole contour; adding the centroid of the bolt hole contour to the reduced distance vector of each point on the bolt hole contour to obtain the new position of each point on the bolt hole contour; the set of the new positions of each point on the bolt hole contour is the new bolt hole contour. Based on the geometric features of the cap stop, the cap stop parameters are calculated using the cap stop point cloud, including the cap stop diameter, center hole diameter, cap groove depth, and wheel core thickness. Based on the geometric features of the bolt holes, the bolt hole parameters are calculated using the bolt hole point cloud, including the bolt hole countersunk diameter, bolt hole diameter, and bolt hole depth. The distance between the bolt hole profile and the cap stop profile is calculated to obtain the positional accuracy between the bolt hole and the cap stop.

Images