Point cloud processing method of line structured light measurement system facing electric drive gear

Abstract

The invention discloses a point cloud processing method of a line structured light measurement system facing an electric drive gear, and belongs to the technical field of gear detection and optical measurement. The method comprises the steps that a system is initialized, and a multi-coordinate system conversion relation is established; acquiring original point clouds of the tooth surface and unifying coordinates; analyzing and mapping the three-dimensional point cloud according to the gear geometric model and reducing the dimension to a two-dimensional tooth profile plane; noise points are removed in the two-dimensional plane based on involute fitting and self-adaptive threshold value recognition; and solving an optimal transformation parameter by adopting an iterative optimization algorithm by taking a minimum point to a theoretical tooth surface normal distance as a target so as to realize high-precision registration of the de-noising point cloud and the theoretical model. According to the method, the specific noise in the gear tooth surface point cloud can be removed in a targeted manner, the real tooth shape is effectively protected, the method has the advantages of being accurate in denoising, stable in registration, high in engineering practicability and the like, and the precision and reliability of electric drive gear line structure light detection are remarkably improved.

Description

Technical Field

[0001] This invention relates to the technical field of gear inspection and optical measurement, and more specifically to a point cloud processing method for a line structured light measurement system for electric drive gears. Background Technology

[0002] Currently, with the development of new energy vehicles and high-end equipment, ensuring the quality of key components such as electric drive gears is crucial. Line structured light scanning technology, with its advantages of non-contact operation, high precision, and high efficiency, is gradually becoming a key method for efficient and non-destructive online measurement of electric drive gears. However, in the actual online structured light scanning process, factors such as complex tooth surfaces causing reflections and viewing angle obstruction result in the acquired tooth surface point cloud containing a large amount of noise and point cloud pose registration errors, thus affecting the subsequent accuracy evaluation of the electric drive gear.

[0003] Several point cloud processing methods exist in the prior art, but they have shortcomings when applied to electric drive gear measurement scenarios. For example, CN120318525A discloses an adaptive joint filtering method based on the relationship between local features and spatial distance for denoising point clouds of autonomous driving LiDAR. Its advantage is that it balances denoising effects for both high-density and sparse point clouds. However, its spatial position relationship adjustment method, when applied to gear point cloud denoising, can lead to tooth profile and tooth shape errors. CN120032043A discloses a method for 3D point cloud matching and reconstruction models for registering 3D point clouds of the underwater well wall portion of a caisson. Its advantage is that it first locates problem points and then performs registration repair, which is more focused than blindly registering the entire cloud. Its disadvantage is that the initial values ​​are strongly dependent on and require scenario assumptions. While this is reasonable for point clouds of the well wall portion of a caisson, it is difficult to apply to gear measurement.

[0004] Therefore, for point cloud data obtained by electric gear line structured light measurement systems, how to efficiently identify and remove point cloud noise after rapid acquisition, and how to accurately register the denoised point cloud with the theoretical model to improve detection accuracy, is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] This invention aims to overcome the shortcomings of existing general point cloud processing methods in dealing with point clouds on the tooth surfaces of electric drive gears, such as inaccurate noise removal, easy damage to features, and insufficient registration accuracy. It provides a point cloud processing method for line structured light measurement systems for electric drive gears, which can effectively remove noise from tooth surface point clouds and achieve high-precision registration.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] This invention provides a point cloud processing method for a line structured light measurement system for electric drive gears, comprising the following steps: W1. System Initialization and Coordinate System Construction: The control measurement system is brought into position, and the coordinate system of the linear structured light sensor is constructed and connected. Intermediate coordinate system of the measurement system and gear coordinate system ; W2. Data Acquisition and Coordinate Unification: Based on the transformation relationship between the three coordinate systems mentioned above, the drive line structured light sensor acquires the original three-dimensional point cloud data of the electric drive gear tooth surface and unifies it to the gear coordinate system. middle; W3, Point Cloud Parsing Mapping and Dimensionality Reduction: In the gear coordinate system Next, the 3D point cloud of the tooth surface is parsed and mapped to a surface coordinate system established based on the tooth surface. Based on the consistent geometric characteristics of the electric drive gears in the tooth width direction, the surface coordinate system is... The 3D point cloud is reduced to a denoised 2D coordinate system consisting of the tooth profile direction and the tooth surface normal direction. In this process, a two-dimensional point set is obtained; W4. Noise Identification and Removal: In the two-dimensional denoising coordinate system In this process, a curve is fitted to the two-dimensional point set of each tooth surface, the distance from each point to the fitted curve is calculated, and an adaptive threshold is set based on the statistical characteristics of the distance. Noise points are removed according to the threshold to obtain the denoised point set. W5. Point Cloud Precise Registration: Construct an objective function using the normal distance from the point cloud in the denoised point set to the theoretical tooth surface as the error metric. Use an optimization algorithm to find the optimal transformation parameters that minimize the objective function. Use the optimal transformation parameters to register the denoised point set with the theoretical tooth surface and output the final 3D point cloud.

[0008] Further, in step W1, the measurement system includes: A line structured light scanning unit includes a line structured light sensor, a chuck for mounting the electric drive gear to be tested, and a torque motor for adjusting the attitude of the sensor. The spatial positioning unit includes a rotary table with the chuck, a rotary table motor for driving the rotary table, a circular grating for feedback of the rotation angle, and a guide rail for mounting the line structured light scanning unit.

[0009] Further, step W2 includes: W2.1 Determine coordinate transformation relationship: Establish coordinate system of line structured light sensor To the gear coordinate system homogeneous transformation matrix It satisfies: ;in, For the coordinate system of the line structured light sensor Transform to the intermediate coordinate system of the measurement system The matrix, For the intermediate coordinate system of the measurement system Transform to gear coordinate system Matrix; W2.2, Constructing the measurement model: In the gear coordinate system Next, establish measurement points. Corresponding point on the theoretical tooth surface Vector relationship model between them: ;in, For measurement points Position vector; To be related to the measurement point Corresponding theoretical points Position vector; For the theoretical point Pointing to the measurement point The deviation vector, the direction of which is perpendicular to the theoretical tooth surface at the theoretical point. The normals at that location are in the same direction; The deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance; for The corresponding unit vector; W2.3, Data Acquisition and Conversion: The rotary table is driven to rotate the electrically driven gear under test, while the line structured light sensor is controlled to continuously acquire raw point cloud data of the gear surface at different rotation angles; for each acquisition moment, based on the real-time rotation angle fed back by the circular grating, the data is converted using the formula... Transform the sensor coordinate system The original point cloud coordinates are uniformly transformed to the gear coordinate system. Below, the initial 3D point cloud after fusion is obtained. .

[0010] Furthermore, in step W3, the parsing mapping specifically involves: Gear coordinate system The three-dimensional point set below Mapped to surface coordinate system The point set below The mapping formula is:

[0011] in, Deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance Let be the base circle radius of the gear to be measured. c is the tooth surface identifier of the electric drive gear under test, with c=-1 for the left tooth surface and c=1 for the right tooth surface; For the first i Measurement points In the gear coordinate system lower edge The coordinate values ​​of the axis; N represents the total number of measurement points.

[0012] Furthermore, in step W3, the dimensionality reduction specifically refers to: Surface coordinate system The three-dimensional point set below Dimensionality reduced to a two-dimensional denoised coordinate system Two-dimensional point set The dimensionality reduction formula is:

[0013] in, Deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance Let be the base circle radius of the gear under test, and c be the tooth surface identifier of the electric drive gear under test. For the left tooth surface, c = -1, and for the right tooth surface, c = 1.

[0014] Furthermore, in step W4, the expression for curve fitting is as follows:

[0015] For point set In a two-dimensional denoised coordinate system The x-coordinate in the middle, The x-axis is The ordinate of the fitted curve. For longitudinal scale parameters, This is the lateral scale parameter. The parameter is the longitudinal offset parameter. , , By the point set The least squares fit is obtained.

[0016] Further, in step W4, the distance from each point to the fitted curve is calculated, and an adaptive threshold is set based on the statistical characteristics of the distance; including: In a two-dimensional denoised coordinate system Longitudinal residuals are used in the middle As a distance metric, calculate each point To the fitted curve European distance :

[0017] Set adaptive threshold ε The formula for identifying outlier noise points is: ,in Two-dimensional point set The mean distance from all points to the fitted curve is given by σ, where σ is the sample standard deviation of the distance and g is the preset threshold coefficient.

[0018] Furthermore, in step W5, the objective function is a residual sum of squares function based on the normal distance:

[0019] Where ζ = (α, ψ, γ, Δx, Δy, Δz) are the registration parameters including rotation and translation. The Euler angle parameters for the registration transformation represent the rotation angles about the coordinate axes X, Y, and Z, respectively. These represent the translation amounts along the X, Y, and Z directions, respectively, during the registration transformation. This represents the point cloud after denoising. Each point in The distance residual vector to the theoretical tooth surface in the normal direction. ; d i ( ζ ) is the first i Each point is measured by parameters ζ The normal distance to the theoretical tooth surface after transformation. i ∈[1,N].

[0020] Furthermore, in step W5, the optimization algorithm is the Gauss-Newton method, which solves the equation iteratively. Update the parameter ζ until the iteration termination condition is met; where, Let r(ζ) be the Jacobian matrix of the residual vector r(ζ) with respect to the parameter ζ at the k-th iteration.

[0021] Further, in step W5, the denoised point set is registered with the theoretical tooth surface using the optimal transformation parameters to output the final 3D point cloud; including: Using the optimal registration parameter ζ, determine the corresponding rotation matrix R(ζ) and translation vector. ; the denoised point set Every point in The coordinate transformation is performed using the following rigid body transformation formula:

[0022] in, For measurement points Position vector; To the measurement point Corresponding theoretical points Position vector; After performing this transformation on all points, the final 3D clean point cloud set is obtained. .

[0023] As can be seen from the above technical solution, compared with the prior art, the present invention provides a point cloud processing method for a line structured light measurement system for electric drive gears. This method includes: system initialization and establishing multi-coordinate system transformation relationships; acquiring the original point cloud of the tooth surface and unifying the coordinates; analytically mapping the three-dimensional point cloud according to the gear geometric model and reducing its dimension to a two-dimensional tooth profile plane; identifying and removing noise points in the two-dimensional plane based on involute fitting and adaptive thresholding; and using an iterative optimization algorithm to solve for the optimal transformation parameters with the goal of minimizing the distance from the point to the theoretical tooth surface normal, achieving high-precision registration between the denoised point cloud and the theoretical model. The present invention can specifically remove specific noise from the gear tooth surface point cloud, effectively protecting the true tooth profile. It has advantages such as accurate noise removal, robust registration, and strong engineering applicability, significantly improving the accuracy and reliability of line structured light detection for electric drive gears.

[0024] The specific technical advantages are as follows: 1. Unified coordinate and normal modeling: By establishing an integrated measurement coordinate system transformation chain and using normal distance to unify the measurement error, a stable and unified mathematical reference framework is provided for subsequent algorithms, enhancing the robustness of the algorithms.

[0025] 2. Dimensionality Reduction and Adaptive Denoising: By utilizing the geometric characteristics of gears, the three-dimensional denoising problem is transformed into a more manageable two-dimensional problem. Combined with statistically based adaptive thresholding to identify noise points, the key geometric features of the gears are preserved to the greatest extent while effectively suppressing noise, achieving boundary-friendly and accurate denoising.

[0026] 3. High-precision robust registration: With the goal of minimizing the normal distance, the method uses iterative algorithms such as Gauss-Newton to solve for the optimal transformation parameters. This method has fast convergence speed, low dependence on initial values, and high registration accuracy, which is significantly better than conventional point-to-point distance-based registration methods.

[0027] 4. Strong engineering applicability: This method has relatively relaxed requirements on the installation and adjustment error of the measurement system and the accuracy of shaft movement. It can adapt well to complex environments such as reflection, obstruction and noise in field measurement. It is easy to integrate with online measurement systems and effectively improves the efficiency and consistency of batch testing of electric drive gears. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0029] Figure 1 This is a flowchart of the point cloud processing method of the line structured light measurement system for electric drive gears according to the present invention. Figure 2 This is a schematic diagram of the line structured light measurement system for electric drive gears according to the present invention; Figure 3 This is a schematic diagram of the coordinate system transformation relationship when using the line structured light sensor 1 to measure the point cloud data of an electric drive gear according to the present invention; Figure 4 This is a schematic diagram of the relative pose relationship of the point cloud transformation coordinate system in a line structured light measurement system for electric drive gears. Figure 5 This is a schematic diagram illustrating the vector relationship between the measurement points and theoretical points of the three-dimensional point cloud information proposed in this invention. Figure 6 This is a schematic diagram illustrating the process of parsing and mapping a point set to a point set in this invention. Figure 7 This is a summary diagram of the process of reducing the dimension of a point set to a two-dimensional point set in this invention; Figure 8 This is a schematic diagram of noise discrimination proposed in this invention; Figure 9 This is a schematic diagram of the registration error model proposed in this invention; Figure 10 This is a schematic diagram of the noise point of the electric drive gear proposed in this invention; Figure 11 The first type of electric drive gear tooth surface point cloud measurement image is not processed by the present invention. Figure 12 This is a first type of electric drive gear tooth surface point cloud measurement image after being processed by the method proposed in this invention; Figure 13 This is a point cloud measurement image of the tooth surface of a second type of electric drive gear with through holes, which has not been processed by this invention. Figure 14 This is a point cloud measurement image of the tooth surface of a second type of electric drive gear with through holes, processed by the method proposed in this invention. Figure 15 A top-down comparison of the point cloud processing effects before and after processing the tooth surface of the first type of electric drive gear. Figure 16 This is a top-down comparison of the point cloud processing effects before and after processing the tooth surface of the second type of electric drive gear with through holes. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] This invention discloses a point cloud processing method for a line structured light measurement system for electric drive gears, referring to... Figure 1 As shown, it includes the following steps: W1. System Initialization and Coordinate System Construction: The control measurement system is brought into position, and the coordinate system of the linear structured light sensor is constructed and connected. Intermediate coordinate system of the measurement system and gear coordinate system ; Among them, the measurement system, such as Figure 2 As shown, it includes: The line structured light scanning unit includes a line structured light sensor 1, a chuck 4 for mounting the electric drive gear 2 to be tested, and a torque motor 3 for adjusting the sensor's attitude. The spatial positioning unit includes a rotary table 6 with a chuck 4, a rotary table motor 7 that drives the rotary table 6, a circular grating 5 for feedback of rotation angle, and a guide rail 8 for mounting the structured light scanning unit.

[0032] During operation, the rotary table 6 and guide rail 8 are homed. The electric drive gear 2 to be tested is clamped onto the rotary table 6 via the chuck 4. The line structured light sensor 1 is adjusted to the preset posture via the torque motor 3 to establish the coordinate system of the line structured light sensor. Intermediate coordinate system of the measurement system and gear coordinate system and establish , , The initial transformation relationship between coordinate systems. The preset attitude refers to the optimized relative angular relationship between the sensor (e.g., torque motor) and the gear tooth surface before the formal measurement begins, established by adjusting these components to obtain high-quality tooth surface point cloud data. This is typically determined and set together with the preset working distance (the optimal distance from the sensor to the tooth surface) to ensure the entire measurement system can obtain usable, high-quality raw point cloud data.

[0033] W2. Data Acquisition and Coordinate Unification: Based on the transformation relationship between the three coordinate systems mentioned above, the drive line structured light sensor acquires the original three-dimensional point cloud data of the electric drive gear tooth surface and unifies it to the gear coordinate system. middle.

[0034] During operation, a measurement model of the electric drive gear 2 under test is established. Based on the measurement model, the rotary table 6 is driven to rotate under the synchronous control of the angle feedback of the circular grating 5. The line structured light sensor 1 continuously collects the original point cloud data of the tooth surface of the electric drive gear 2 under test at different angles according to the preset posture. The data is then transmitted through the coordinate system of the line structured light sensor. Intermediate coordinate system of the measurement system and gear coordinate system The homogeneous transformation relationship between them uniformly transforms the raw point cloud data output by the line structured light sensor 1 to the measured gear coordinate system. middle.

[0035] W3, Point Cloud Parsing Mapping and Dimensionality Reduction: In the gear coordinate system Next, the 3D point cloud of the tooth surface is parsed and mapped to a surface coordinate system established based on the tooth surface. Based on the consistent geometric characteristics of the electric drive gears in the tooth width direction, the surface coordinate system is... The 3D point cloud is reduced to a denoised 2D coordinate system consisting of the tooth profile direction and the tooth surface normal direction. In this process, a two-dimensional point set is obtained.

[0036] This step involves introducing a mapping from spatial coordinate points to a surface coordinate system, first constructing a surface coordinate system on the tooth surface of the electric drive gear 2 under test. The transformation formula between spatial coordinate points and gear tooth surface structure is obtained by chain rule differentiation. Based on the transformation formula, the gear coordinate system is transformed. The surface coordinate system is obtained by mapping the three-dimensional point cloud. The three-dimensional point set, based on the characteristic that the geometric features of the electric drive gear tooth surface are consistent in the tooth width direction, is only in the surface coordinate system. Construct a two-dimensional noise-reducing coordinate system along the tooth profile direction and the tooth surface normal direction. using the gear coordinate system Using the Z-axis as the dimension reduction direction, the surface coordinate system Transforming a 3D point set into a 2D denoised coordinate system A two-dimensional point set.

[0037] W4. Noise Identification and Removal: In the two-dimensional denoising coordinate system In this process, a curve is fitted to the two-dimensional point set of each tooth surface, the distance from each point to the fitted curve is calculated, and an adaptive threshold is set based on the statistical characteristics of the distance. Noise points are removed according to the threshold to obtain the denoised point set. This step is based on the surface coordinate system. For each tooth of the electric drive gear 2 under test, a two-dimensional set of points to be processed is extracted from the tooth surface. The least squares fitting is performed on the two-dimensional set of points to be processed to obtain the fitted curve equation. The Euclidean distance from each point in the two-dimensional set of points to be processed to the fitted curve equation is calculated, and a threshold criterion is given to complete the noise identification and removal.

[0038] W5. Precise Point Cloud Registration: An objective function is constructed using the normal distance from the denoised point cloud to the theoretical tooth surface as the error metric. An optimization algorithm is used to find the optimal transformation parameters that minimize the objective function. These optimal transformation parameters are then used to register the denoised point cloud with the theoretical tooth surface, outputting the final 3D point cloud. In this step, the normal distance in the established measurement model of the electric drive gear 2 is used as the error metric. A nonlinear least squares objective is constructed by minimizing the sum of the normal distances from points to the tooth surface. The Gauss-Newton iterative algorithm is used to solve for the optimal registration parameters, obtaining the optimal transformation matrix. This allows for precise registration of the electric drive gear tooth surface point cloud with the theoretical tooth surface, resulting in a clean point cloud.

[0039] This method establishes a measurement model for the normal distance of the tooth profile of an electric drive gear, reduces the dimensionality of the three-dimensional point cloud data of the electric drive gear to two-dimensional tooth profile normal distance, thereby adaptively identifying and eliminating optical noise, and then iteratively solves the optimal transformation matrix for accurate registration.

[0040] The five steps of the present invention will now be explained in detail: W1: Measurement system repositioning and establishment of point cloud coordinate system; W1.1: Measurement system initialization; Build a line structured light measurement system for electric drive gears, such as Figure 2 As shown, the system includes a line structured light scanning unit and a spatial positioning unit. A zero-return operation is performed on the guide rail 8 of the spatial positioning unit. The turntable motor 7 is started, causing the rotary table 6 to rotate to the zero angle of the circular grating 5, and this zero angle is recorded as an angle reference. The electric drive gear 2 to be tested in the line structured light scanning unit is mounted on the chuck 4 and clamped. The rotation axis of the electric drive gear 2 to be tested and the rotary table 6 are adjusted to be coaxial. The torque motor 3 is started to adjust the relative position of the line structured light sensor 1 and the electric drive gear 2 to be tested, so that the line structured light sensor 1 is at a preset working distance and preset posture to meet the requirements for tooth surface point cloud acquisition.

[0041] W1.2: Establish a point cloud transformation coordinate system; like Figure 3 As shown, three coordinate systems are established: a point cloud transformation coordinate system is established around the electric drive gear 2 under test, and the coordinate system of the line structured light sensor is established. , recorded as Intermediate coordinate system of the measurement system , recorded as Gear coordinate system , recorded as Among them, the coordinate system of the line structured light sensor Fixed to the line structured light sensor 1, used to represent the original sampling points output by the line structured light sensor 1; intermediate coordinate system of the measurement system. Fixed between the line structured light scanning unit and the spatial positioning unit, it is used to uniformly describe the positional relationship between the line structured light sensor 1 and the theoretical gear surface; gear coordinate system Fixed to the electric drive gear 2 under test and rotating synchronously with the rotary table 6, the gear coordinate system Relative measurement system intermediate coordinate system The rotation angle is determined by the angle output of the circular grating 5, and is used for subsequent point cloud coordinate transformation and tooth surface point cloud fusion reconstruction.

[0042] W2: Building the measurement model and collecting data; W2.1: Determine the coordinate transformation relationship; like Figure 3 As shown, the coordinate system of the line structured light sensor is... coordinates Unified transformation to gear coordinate system coordinates The transformation relationship is shown in formula (1): (1) Line structured light sensor coordinate system To the gear coordinate system The homogeneous transformation matrix is ​​denoted as It satisfies formula (2): (2) in, coordinate system of line structured light sensor To the intermediate coordinate system of the measurement system The transformation matrix, For the intermediate coordinate system of the measurement system To the gear coordinate system The transformation matrix, and As shown in the formula below: (3) (4) In the formula, for , for , for , for , for , for ;like Figure 4 As shown, For the coordinate system of the line structured light sensor Around Roll angle, For the coordinate system of the line structured light sensor Around pitch angle, For the coordinate system of the line structured light sensor Around The yaw angle, h , j , k The coordinate system of the structured light sensor is respectively Relative measurement system intermediate coordinate system Three-axis The translation distance, The electric drive gear 2 under test is located in the intermediate coordinate system of the measurement system. Middle winding The rotation angle of the shaft, where the gear coordinate system of Axis and intermediate coordinate system of measurement system of By setting the axes coaxially, the coordinate system of the electric drive gear under test can be obtained. The three-dimensional point cloud information below is denoted as a point set. , As shown in equation (5): (5) N represents the total number of measurement points.

[0043] W2.2: Constructing a measurement model; like Figure 5 As shown, a measurement model with normal distance as the core is established: the electric drive gear 2 under test is in the gear coordinate system. Measurement points in the following 3D point cloud information With theoretical point There exists a vector relationship between the measurement points, that is: The position vector is equal to its corresponding theoretical point The position vector is added to the normal distance vector; a measurement model is then established based on this, as shown in the following formula: (6) For measurement points position vector, To be related to this measurement point Corresponding theoretical points position vector, For the theoretical point Pointing to the measurement point The deviation vector, the direction of which is perpendicular to the theoretical tooth surface at the theoretical point. The normals at that location are in the same direction; The deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance; for The corresponding unit vector; W2.3: Collect point cloud data of the tooth surface of the electric drive gear; Maintaining the line structured light sensor 1 at a preset working distance and preset posture relative to the electric drive gear 2 under test, the high-speed synchronous data acquisition system is activated; the angle feedback signal of the circular grating 5 is input. θ As a synchronous sampling reference, the turntable motor 7 is started to drive the rotary table 6 to rotate, which in turn drives the electric drive gear 2 under test to rotate around the rotation axis. During the rotation process, the line structured light sensor 1 synchronously collects data at various angles. θ The corresponding tooth surface point cloud data, and the tooth surface point cloud data with the corresponding angle θ Related records and storage; Rotary table 6 completed. After rotation and stopping, a complete point cloud data sequence from the tooth tip to the tooth root of the same tooth surface of the electric drive gear 2 under test is obtained. Based on the radii of the tooth tip and tooth root to the center of the gear, the point cloud data of the tooth tip and tooth root are removed, and the tooth surface point cloud data is obtained as the input data for subsequent steps. The preset working distance is a distance from the sensor to the tooth surface, similar to the preset posture. The combination of the two can ensure the quality of data acquisition.

[0044] W3: Analytical mapping and dimensionality reduction representation of point cloud data of electric drive gear tooth surfaces: W3.1: Parsing and mapping of point cloud data for the tooth surface of electric drive gears: like Figure 6 As shown, a three-dimensional diagram demonstrates the gear coordinate system. The original point cloud below Mapped to the tooth surface coordinate system through analytical calculation get The process.

[0045] Construct a surface coordinate system on the gear tooth surface for ,in Along the direction of the tooth surface normal, Along the tooth profile direction, Along the tooth width direction, the gear coordinate system next point set point Mapped to surface coordinate system The point set below To achieve parameterized representation of tooth surface point cloud data, This is denoted as equation (7): (7) The mapping process is shown in formula (8): (8) in, It is the radius of the base circle of the electric drive gear under test. , The values ​​are taken from the left and right tooth surfaces of the electric drive gear 2 under test, using the gear coordinate system. of The direction is positive, left tooth surface =-1, right tooth surface =1, For the first i Measurement points In the gear coordinate system lower edge The coordinate values ​​of the axis are shown in formula (9): (9) If a tooth on the electric drive gear 2 under test is designated as the starting tooth, then... T It is the second electric drive gear under test. i The tooth number of each tooth. Indicates the first i Measurement points The azimuth angle on the end face of the electric drive gear 2 under test. The total number of teeth of the electric drive gear 2 under test. As shown in formula (10): (10) This is the pressure angle of the end face of the second index circle of the electric drive gear under test. This completes the point cloud data set of the electric drive gear tooth surface. Point set The parsing mapping.

[0046] W3.2: Dimensionality Reduction Representation of Point Cloud Data for Electric Drive Gear Tooth Surfaces: like Figure 7 As shown, the parallel arrangement of three coordinate systems emphasizes that point m lies in the three-dimensional gear coordinate system. Three-dimensional surface coordinate system and two-dimensional denoised coordinate system The coordinate transformation relationship below.

[0047] Taking advantage of the consistent characteristics of gear tooth width direction, in the tooth profile direction of the electric drive gear 2 under test... With the direction of the tooth surface normal Construct a two-dimensional denoised coordinate system , Recorded as , surface coordinate system midpoint set Using the gear coordinate system The Z-axis is the direction of dimensionality reduction, which will reduce the point set... Transform to a set of points in a 2D denoised coordinate system dot set It is denoted as Equation (11): (11) The conversion process is shown in formula (12): (12) The physical meaning is consistent with formula (6). Deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance is thus obtained, and the dimensionality reduction representation of the point cloud data of the tooth surface of the electric drive gear under test is completed. Dimensionality reduction greatly simplifies the data complexity.

[0048] W4: Identification and removal of noise in point cloud data of electric drive gear teeth; W4.1: Fitting the equation of a two-dimensional point cloud curve: In the two-dimensional denoising coordinate system established by W3 In the process, a set of points to be processed is extracted from each tooth surface of the electric drive gear 2 under test. Based on point sets The curve function is obtained by using the least squares criterion, and the curve equation is fitted. As shown in equation (13): (13) t For point set In a two-dimensional denoised coordinate system The x-coordinate in the middle, The x-axis is t The ordinate of the fitted curve is given by B, the vertical scale parameter is given by B, the horizontal scale parameter is given by C, and the vertical bias parameter is given by D. These parameters B, C, and D are obtained by applying a set of points... The least squares fit is obtained.

[0049] W4.2: Set threshold to identify noise: Since the noise on the tooth surface of the electric drive gear 2 under test is mainly manifested as a deviation along the normal direction of the tooth surface, in the two-dimensional noise-reducing coordinate system... Longitudinal residuals are used l As a distance metric, such as Figure 8The diagram shown illustrates the principle of two-dimensional planar noise recognition, within a two-dimensional denoising coordinate system. In the diagram, the scatter points represent the dimensionality-reduced data, and the curve represents the fitted theoretical tooth profile. The vertical distance (residual) from each point to the curve is calculated and compared with a set threshold. Comparison is used to determine whether it is noise.

[0050] Specifically, each point is calculated. To the fitted curve L European distance , As shown in formula (14): (14) Set threshold The threshold calculation formula for identifying outlier noise points is shown in (15): (15) For point set To the fitted curve The mean distance, The distance to the sample standard deviation is calculated as follows: (16) (17) n is a point set The number of points, g The preset threshold coefficient can be set, for example, the value range is 0.01-0.05.

[0051] W4.3: Noise Removal like Figure 8 As shown in the figure, "noise" and the decision inequality are clearly marked. > (Noise) and ≤ (Valid points).

[0052] Based on distance l With threshold Noise removal is performed on the comparison results: when At that time, the point Identify as noise points and remove them; when At that time, reserve point This method can effectively identify and remove outlier noise caused by double reflections, etc.

[0053] W5: Registration of the denoised point cloud with the theoretical tooth surface; W5.1: Establishing error metrics and objective functions; Let the denoised point cloud set be denoted as , The normal distance in W2.2 As an error metric, using point sets The objective is to minimize the sum of squared residuals in the distance to the theoretical tooth surface normal direction. A nonlinear least squares objective function is constructed as shown in equation (18): (18) Where ζ = (α, ψ, γ, Δx, Δy, Δz) are the registration parameters including rotation and translation. The Euler angle parameters for the registration transformation represent the rotation angles about the coordinate axes X, Y, and Z, respectively. These represent the translation amounts along the X, Y, and Z directions, respectively, during the registration transformation. Point cloud Each point in The distance residual vector to the theoretical tooth surface in the normal direction. , For registration parameters, ,in, Euler angles, It is a translation vector; d i ( ζ ) represents the parameters of the i-th point. ζ The normal distance from the transformed tooth surface to the theoretical tooth surface, i∈[1,N].

[0054] W5.2: Iterative solution and parameter update; Figure 9 It shows the state before and after registration. These are the measured points after noise reduction. These are the corresponding points on the theoretical tooth surface. A set of rotation and translation parameters are solved using the Gauss-Newton method optimization algorithm, ensuring that after transformation of all measured points, their... The corresponding theoretical point along Vector direction to measurement point straight-line distance The sum of squares is the smallest.

[0055] Specifically, the Gauss-Newton method is used to... Perform iterative minimization, when the... k In the next iteration, the residual vector is linearized to the first order to obtain the parameter increment. The linear equation that is satisfied is shown in equation (19): (19) in, , Let be the Jacobian matrix, as shown in equation (20): (20) Will Substituting into equation (18), for Finding the minimum value yields the Gauss-Newton incremental equation, as shown in equation (21): (twenty one) Solve for parameter increments Then, the registration parameters are calculated according to the following update rules: (twenty two) When parameter increment satisfy or the gradient norm is sufficiently small or reaching the maximum number of iterations The iteration terminates at time, where , For the preset threshold, The maximum number of iterations is set, and the optimal registration parameters are output upon termination. .

[0056] W5.3: Register and output clean point clouds; Using optimal parameters The final rigid body transformation is obtained. and ,Depend on A defined rigid body transformation will remove noise. The transformation is shown in equation (23): (twenty three) in, Let be a rotation matrix. The translation vector will be used to transform the denoised point set. Each point in the middle is subjected to coordinate transformation according to the final rigid body transformation formula (23) to obtain the point set after registration with the theoretical tooth surface. This completes the clean point cloud output.

[0057] Figure 10 It is the raw, unprocessed gear point cloud. You can see that the point cloud is messy and contains a lot of discrete noise (similar to snowflakes).

[0058] like Figure 11-16 As shown, the method of the present invention is applied to point cloud processing of two different types of electric drive gears.

[0059] Group 1: Global effect comparison of standard gears: like Figure 11This is a point cloud measurement image of the tooth surface of the first type of electric drive gear. This is the raw scanned point cloud without any processing. As can be seen, there are a large number of discrete, messy, and randomly distributed noise points around the three-dimensional shape of the gear, like "snowflakes" or "dust" spreading throughout the space. These noise points severely interfere with the identification of the gear's true contour, making the boundaries of key areas such as the tooth surface and tooth grooves unclear.

[0060] like Figure 12 As shown, this is a point cloud measurement image of the first type of electric drive gear tooth surface after processing by the method proposed in this invention, and... Figure 11 In stark contrast, all discrete noise has been completely eliminated. The remaining point cloud is clear, smooth, and structurally complete, accurately outlining the three-dimensional geometry of each tooth, tooth groove, and end face of the gear. The point cloud data is of extremely high quality and can be directly used for high-precision gear error evaluation.

[0061] Group 2: Comparison of challenging cases involving gears with through holes: Figure 13 The image shows the point cloud measurement of the tooth surface of the second type of electric drive gear with a through hole; it is the original point cloud of a gear with a central through hole and spokes. The noise problem is equally serious, and the noise points are more concentrated inside and at the edge of the through hole due to multiple reflections and occlusions, forming obvious noise bands that almost completely obscure the true structure of the through hole.

[0062] like Figure 14 As shown, this is a second point cloud measurement image of the tooth surface of an electric drive gear with through holes, processed by the method proposed in this invention. The effect of processing with the method of this invention is equally amazing: the annoying dense noise inside and around the through hole is completely removed, and the cylindrical surface of the through hole is clearly visible; all features such as the spoke structure and tooth surface are cleanly presented. The entire point cloud model is very "clean" and "robust".

[0063] This demonstrates that the method of the present invention is not only applicable to standard gears, but also effective for gears with complex internal cavity structures such as holes and slots, showcasing the method's strong adaptability and robustness.

[0064] Third group: Detail comparison from a top-down perspective: Figure 15 The first type of electric drive gear shows a top-down comparison of the point cloud before and after processing, comparing the processing effects from the gear axis (top view). The outline on the left before processing appears rough and discontinuous due to the presence of peripheral noise, and the outer edge of the outline is covered with scattered points. The tooth tip circle and tooth root circle on the right after processing have smooth, continuous, and regular outlines.

[0065] From a two-dimensional projection perspective, the effectiveness of the method of this invention in removing external noise and purifying boundaries is highlighted. This demonstrates that the method can accurately remove external pseudo-points while protecting the actual shape boundary of the gear.

[0066] Figure 16 The second type of electric drive gear with through holes is shown in the top-view comparison of point clouds before and after processing. Also a top-view comparison, this image focuses on gears with through holes. On the left, before processing, the through-hole area is covered by a chaotic cloud of noise, making it impossible to discern the shape of the hole. On the right, after processing, the circular boundary of the through hole is extremely clear and regular, and the spoke structure is distinct.

[0067] The method of this invention can effectively handle noise at the boundaries of complex internal structures. Figure 15 , Figure 16 Further evidence shows that the denoised point cloud has a clear outline and fits well with the theoretical tooth surface, verifying the excellent effect of the method of the present invention in denoising and registration.

[0068] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0069] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A point cloud processing method for a line structured light measurement system for electric drive gears, characterized in that, Includes the following steps: W1. System Initialization and Coordinate System Construction: The control measurement system is brought into position, and the coordinate system of the linear structured light sensor is constructed and connected. Intermediate coordinate system of the measurement system and gear coordinate system ; W2. Data Acquisition and Coordinate Unification: Based on the transformation relationship between the three coordinate systems mentioned above, the drive line structured light sensor acquires the original three-dimensional point cloud data of the electric drive gear tooth surface and unifies it to the gear coordinate system. middle; W3, Point Cloud Parsing Mapping and Dimensionality Reduction: In the gear coordinate system Next, the 3D point cloud of the tooth surface is parsed and mapped to a surface coordinate system established based on the tooth surface. Based on the consistent geometric characteristics of the electric drive gears in the tooth width direction, the surface coordinate system is... The 3D point cloud is reduced to a denoised 2D coordinate system consisting of the tooth profile direction and the tooth surface normal direction. In this process, a two-dimensional point set is obtained; W4. Noise Identification and Removal: In the two-dimensional denoising coordinate system In this process, a curve is fitted to the two-dimensional point set of each tooth surface, the distance from each point to the fitted curve is calculated, and an adaptive threshold is set based on the statistical characteristics of the distance. Noise points are removed according to the threshold to obtain the denoised point set. W5. Point Cloud Precise Registration: Construct an objective function using the normal distance from the point cloud in the denoised point set to the theoretical tooth surface as the error metric. Use an optimization algorithm to find the optimal transformation parameters that minimize the objective function. Use the optimal transformation parameters to register the denoised point set with the theoretical tooth surface and output the final 3D point cloud.

2. The method as described in claim 1, characterized in that, In step W1, the measurement system includes: The line structured light scanning unit includes a line structured light sensor (1), a chuck (4) for mounting the electric drive gear (2) to be tested, and a torque motor (3) for adjusting the attitude of the sensor. The spatial positioning unit includes a rotary table (6) with the chuck (4), a rotary table motor (7) for driving the rotary table (6), a circular grating (5) for feedback of rotation angle, and a guide rail (8) for mounting the line structured light scanning unit.

3. The method as described in claim 2, characterized in that, Step W2 includes: W2.1 Determine coordinate transformation relationship: Establish coordinate system of line structured light sensor To the gear coordinate system homogeneous transformation matrix It satisfies: ;in, For the coordinate system of the line structured light sensor Transform to the intermediate coordinate system of the measurement system The matrix, For the intermediate coordinate system of the measurement system Transform to gear coordinate system Matrix; W2.2, Constructing the measurement model: In the gear coordinate system Next, establish measurement points. Corresponding point on the theoretical tooth surface Vector relationship model between them: ;in, For measurement points Position vector; To the measurement point Corresponding theoretical points Position vector; For the theoretical point Pointing to the measurement point The deviation vector, the direction of which is perpendicular to the theoretical tooth surface at the theoretical point. The normals at that location are in the same direction; The deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance; for The corresponding unit vector; W2.3, Acquisition and Conversion of Point Cloud Data: The rotary table (6) drives the electric drive gear (2) under test to rotate, while controlling the line structured light sensor (1) to continuously acquire the original point cloud data of the tooth surface at different rotation angles; for each acquisition moment, based on the real-time rotation angle fed back by the circular grating (5), the formula is used to convert the point cloud data into point cloud data. Transform the sensor coordinate system The original point cloud coordinates are uniformly transformed to the gear coordinate system. Below, the initial 3D point cloud set after fusion is obtained. .

4. The method as described in claim 1, characterized in that, In step W3, the parsing mapping specifically involves: Gear coordinate system The three-dimensional point set below Mapped to surface coordinate system The point set below The mapping formula is: in, Deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance Let be the base circle radius of the gear to be measured. c is the tooth surface identifier of the electric drive gear under test, with c=-1 for the left tooth surface and c=1 for the right tooth surface; For the first i Measurement points In the gear coordinate system lower edge The coordinate values ​​of the axis; N represents the total number of measurement points.

5. The method as described in claim 4, characterized in that, In step W3, the dimensionality reduction specifically refers to: Surface coordinate system The three-dimensional point set below Dimensionality reduced to a two-dimensional denoised coordinate system Two-dimensional point set The dimensionality reduction formula is: in, Deviation vector The modulus, representing the length of the theoretical point. along Vector direction to measurement point The straight-line distance Where is the base circle radius of the gear to be tested, and c is the tooth surface identifier of the electric drive gear to be tested, with c=-1 for the left tooth surface and c=1 for the right tooth surface.

6. The method as described in claim 5, characterized in that, In step W4, the expression for curve fitting is as follows: For point set In a two-dimensional denoised coordinate system The x-coordinate in the middle, The x-axis is The ordinate of the fitted curve. For longitudinal scale parameters, This is the lateral scale parameter. The parameter is the longitudinal offset parameter. , , By the point set The least squares fit is obtained.

7. The method as described in claim 6, characterized in that, In step W4, the distance from each point to the fitted curve is calculated, and an adaptive threshold is set based on the statistical characteristics of the distance; this includes: In a two-dimensional denoised coordinate system Longitudinal residuals are used in the middle As a distance metric, calculate each point To the fitted curve European distance : Set adaptive threshold ε The formula for identifying outlier noise points is as follows: ,in Two-dimensional point set The mean distance from all points to the fitted curve is given by σ, where σ is the sample standard deviation of the distance and g is the preset threshold coefficient.

8. The method as described in claim 1, characterized in that, In step W5, the objective function is a residual sum of squares function based on the normal distance: Where ζ = (α, ψ, γ, Δx, Δy, Δz) are the registration parameters including rotation and translation. The Euler angle parameters for the registration transformation represent the rotation angles about the coordinate axes X, Y, and Z, respectively. These represent the translation amounts along the X, Y, and Z directions, respectively, during the registration transformation. This represents the point cloud after denoising. Each point in The distance residual vector to the theoretical tooth surface in the normal direction. ; d i ( ζ ) is the first i Each point is measured by parameters ζ The normal distance to the theoretical tooth surface after transformation. i ∈[1,N].

9. The method according to claim 8, characterized in that, In step W5, the optimization algorithm is the Gauss-Newton method, which solves the equation iteratively. Update the parameter ζ until the iteration termination condition is met; where, For the first k The Jacobian matrix of the residual vector r(ζ) with respect to the parameter ζ at the next iteration.

10. The method according to claim 9, characterized in that, In step W5, the denoised point set is registered with the theoretical tooth surface using the optimal transformation parameters to output the final 3D point cloud; including: Using the optimal registration parameter ζ, determine the corresponding rotation matrix R(ζ) and translation vector. ; the denoised point set Every point in The coordinate transformation is performed using the following rigid body transformation formula: in, For measurement points Position vector; To the measurement point Corresponding theoretical points Position vector; After performing this transformation on all points, the final 3D clean point cloud set is obtained. .

Images