A line structured light type gear rotation scanning system and a rotation axis calibration method thereof

Abstract

The application discloses a line structure light type gear rotation scanning system and a rotation axis calibration method thereof, three coordinate systems of a camera coordinate system, a base coordinate system and a workpiece coordinate system and conversion relations are established in the scanning system; initial calibration of a hand-eye matrix is completed by using a joint arm measuring machine and point cloud processing software according to changes of incident structure light in the workpiece and the camera coordinate system; comprehensive optimization of whole rotation axis parameters is completed by using an initial calibration output matrix and an ICP algorithm; the method has the advantages of fast calibration speed, high efficiency, simple operation and good optimization effect.

Description

Technical Field

[0001] This invention relates to the field of visual inspection system calibration technology, specifically to a line structured light gear rotary scanning system and its rotary axis calibration method. Background Technology

[0002] In recent years, numerous scholars both domestically and internationally have conducted extensive research on 3D tooth surface measurement. Line structured light measurement, as a commonly used non-contact 3D point cloud measurement method, can quickly obtain the tooth surface morphology of parts. Line structured light measurement technology projects uniform light rays onto the surface of the object being measured, utilizing the deformation and displacement of the light rays on the object's surface to acquire 3D shape data. The measurement system captures images of the light deformation through a camera in a sensor and uses image processing algorithms to analyze the changes in the light stripes, thereby accurately reconstructing the 3D shape of the object. Due to its high speed, high precision, and anti-interference characteristics, it has been widely used in 3D measurement.

[0003] Line-structured light (SCL) non-contact gear measurement technology has become a research hotspot in the global non-contact measurement field. However, compared with foreign research, its development in China lags behind. The reasons are twofold: firstly, the mainstream method used domestically is still contact measurement, leading to a lack of investment in this area. Secondly, my country is relatively behind the United States, Germany, and Japan in the field of optoelectronic information technology. my country lacks a mature industrial chain for the production and processing of SCL probes; relying entirely on imported equipment would be prohibitively expensive for most manufacturers, hindering its widespread adoption domestically. In addition, there are several issues that must be addressed in SCL 3D measurement, including coordinate relationship calibration and gear triangular mesh reconstruction in SCL 3D gear measurement systems. Therefore, further research and improvement of SCL non-contact gear measurement technology are necessary.

[0004] Existing technologies, such as the calibration scheme disclosed in Wang Tao's dissertation "Research on Gear 3D Measurement Method Based on Line Structured Light" at Beijing University of Technology, involve: firstly, determining the measurement reference point through complex analysis using a mathematical model; secondly, obtaining multiple sets of non-collinear calibration points through multiple scans; and finally, calibrating the rotating shaft by satisfying the external parameter calculations of the relative motion model. He Wantao and Ma Heyao's calibration scheme, "Precision Calibration Method for Rotating Shafts in Non-Contact Optical Measurement of Aeronautical Blades," involves: obtaining the point cloud of the cross-sectional profile of a standard sphere using a translational scanning method; collecting data multiple times by rotating the turntable; fitting the sphere center using singular value decomposition; and finally solving the equation of the sphere center trajectory plane and the rotating shaft normal vector using the least squares method. The intersection point of the rotating shaft and the trajectory plane is then determined, which is the rotation shaft center. All of these existing methods for calibrating rotating shaft parameters suffer from cumbersome processes and low efficiency. Therefore, a new method must be proposed to address the shortcomings of existing technologies. Summary of the Invention

[0005] 1. The technical problem to be solved:

[0006] To address the aforementioned technical problems, this invention provides a line structured light gear rotation scanning system and its rotation axis calibration method. The system comprises a rotary table and multiple structured light sensors, enabling omnidirectional scanning of complex parts. This allows the sensor's field of view to fully cover any part of the complex workpiece. Point clouds acquired from various angles are rotated and stitched together to complete a three-dimensional reconstruction of the blade's morphology. The effectiveness of the point cloud rotation stitching depends on the calibration accuracy of the rotation axis.

[0007] 2. Technical Solution:

[0008] A line structured light gear-like rotary scanning system is characterized by comprising multiple structured light sensors, a multi-degree-of-freedom optical adjustment frame, a precision angular displacement stage, an articulated arm measuring machine, and a central turntable. The multiple structured light sensors are mounted on the multi-degree-of-freedom optical adjustment frame and the angular displacement stage to acquire point cloud information of the workpiece from different perspectives. The multi-degree-of-freedom optical adjustment frame adjusts the structured light sensors to their positions on the XYZ axes, and the angular displacement stage fine-tunes the angle of the light sensors to obtain the optimal scanning pose. The laser scanning probe of the articulated arm measuring machine acquires the STL model of the workpiece, which is used to calibrate the accuracy of the point cloud acquired by the system. The central turntable is a disc-type turntable, using its own central axis as the motion reference, and a drive mechanism enables it to rotate around the central turntable axis. The workpiece is fixed to a preset positioning area on the central turntable surface and rotates synchronously with the rotation of the central turntable, thereby enabling the line structured light sensors to continuously scan and acquire data from the surface contour within the corresponding perspective during the workpiece's movement.

[0009] Furthermore, the end effector of the articulated arm measuring machine integrates a standardized connection structure, which has a modular interface function and can realize detachable fixation and precise positioning of the scanning probe.

[0010] A method for calibrating the rotating shaft of a line structured light gear-type rotary scanning system includes the following steps:

[0011] Step 1: Set up a line structured light gear rotary scanning system; place the standard part to be inspected on the central turntable; adjust the positions of multiple structured light sensors so that the sum of the viewing angles of the multiple structured light sensors can cover the entire outline of the object to be measured; install a laser scanning probe at the end of the articulated arm measurement;

[0012] Step 2: Determine the base coordinate system of the line structure light gear rotary scanning system; take the intersection of the central rotation axis of the central turntable and the table surface as the origin of the base coordinate system, the Z-axis direction is the same as the rotation axis of the central turntable, and the plane formed by the X and Y axes is the plane of the central turntable table surface, which does not rotate with the rotation of the turntable.

[0013] Step 3: Determine the corresponding camera coordinate system for each structured light sensor: Select the camera optical center or the imaging center of the image plane of the sensor as the origin of the coordinate system. Its X and Z axes together constitute the light plane of the line structured light sensor. The Z-axis direction is parallel to and opposite to the line structured light projection direction, and the Y-axis is perpendicular to the light plane according to the right-hand rule.

[0014] Step 4: Control the articulated arm measuring machine to scan the outer surface of the standard part to obtain its three-dimensional geometric model and output it in STL format to obtain a triangular mesh STL model; based on the triangular mesh STL model, use area weighting and point spacing constraints to generate a three-dimensional point cloud set that meets the preset density and distribution characteristics; this three-dimensional point cloud set is the calibrated target point cloud;

[0015] Step 5: Analyze all structured light sensors to obtain the initial transformation matrix from the camera coordinate system to the base coordinate system. The unit rotation angle is preset, and the turntable is controlled to drive the standard part to rotate synchronously for one revolution. The structure light sensor at different position collects the point cloud corresponding to each unit rotation angle. All the point clouds acquired by the structure light sensor at the same position in this process constitute the complete long strip point cloud of the standard part's outline. Based on the transformation matrix M between the base coordinate system and the camera coordinate system, each long strip point cloud is rotated and stitched together to obtain the rotated scan point cloud of the outline of the part under test.

[0016] Step 6: Based on ICP point cloud registration, define the rotated scan point cloud as the source point cloud, iteratively optimize the relative position and orientation between the target point cloud and the source point cloud, so that the distance between the corresponding point clouds gradually decreases, thereby minimizing the error and obtaining the rotation axis parameters of the current state.

[0017] Furthermore, the standard part is a geometric model of a polyhedron, and its shape is provided with the geometric shapes of planes, trapezoids, and holes.

[0018] Furthermore, in step four, the triangular mesh STL model is converted from a mesh to a point cloud using reverse modeling software.

[0019] Furthermore, step five specifically includes:

[0020] S51: Taking one of the structured light sensors as an example, the first frame point cloud obtained through reverse processing software is matched to the model position corresponding to the workpiece position where the incident light is located, thus obtaining the global translation matrix of the first frame point cloud, i.e., the initial transformation matrix. ;

[0021] S52: Take the unit vector on the Z-axis of the base coordinate system. From the initial transformation matrix The vector n is obtained by inverse transformation to the camera coordinate system; the starting point of vector n is translated to the origin of the camera coordinate system, and the translation parameter is... Then the translation transformation matrix T is as follows:

[0022] (1);

[0023] In the above formula, t x t y t z These represent the translation parameters on the x, y, and z coordinate axes respectively during this process;

[0024] S53: Rotate sequentially around the X, Y, and Z axes of its own coordinate system from the initial state;

[0025] When rotating around the X-axis by an angle When, its corresponding rotation matrix R x ( ) is represented as:

[0026] (2);

[0027] When rotating around the Y-axis by an angle β, since counterclockwise rotation is defined as positive in a right-handed coordinate system, the corresponding rotation angle for clockwise rotation should be negative, and the corresponding rotation matrix R... y (-β) is as follows:

[0028] (3);

[0029] When rotating around the Z-axis by an angle γ, the corresponding rotation matrix R z (γ) is represented as:

[0030] (4);

[0031] Combining equations (2), (3), and (4), we obtain the rotation transformation matrix R. xyz As shown in the following formula:

[0032] (5);

[0033] In the above formula, β is the angle between the direction vector n of the current frame point cloud and the X-axis in the XOZ plane; β is the angle between the direction vector n of the current frame point cloud and the Y-axis in the YOZ plane; γ represents the rotation angle of the corresponding axis cylinder of the current frame point cloud around the Z-axis;

[0034] S54: Equation (5) is the transformation from the base coordinate system to the camera coordinate system. Therefore, the transformation matrix M can be obtained by its inverse process, that is:

[0035] (6);

[0036] S55: As in steps S51-54, until the transformation matrix M corresponding to all structured light sensors is obtained, and the transformation matrix M is used to realize the rotation and stitching of all strip-shaped point clouds.

[0037] Furthermore, step six involves designing and implementing a point cloud registration algorithm based on the ICP algorithm, and employing a grid grouping strategy and iterative optimization method for parameter solving and accelerated computation; the specific process includes:

[0038] S61: The preset rotation axis includes 11 parameters as shown in the following formula, and its parameter vector expression par is as follows:

[0039] (7);

[0040] In the above formula, r x， r y， r z These represent the rotation parameters for the corresponding coordinate axes; p x p y p z These are the coordinates of the starting point of the rotation axis corresponding to the coordinate axis; these 11 parameters can fully reflect the transformation matrix M;

[0041] S62: Perform denoising and outlier removal preprocessing on both the target point cloud and the source point cloud; perform downsampling on the preprocessed point cloud; and then use the processed target point cloud... Representation, source cloud usage express;

[0042] S63: Divide the target point cloud space into multiple small cells, each cell containing a local subset of the target point set. The initial grid step size S is:

[0043] (8);

[0044] In the above formula, h is a parameter used to scale the point cloud, and its size is set according to the total number of points; and These represent the maximum and minimum values ​​of the target point cloud along its x, y, and z axes, respectively; a point cloud set is taken from the target point cloud set. Find the corresponding point cloud set in the source point cloud set. , making Minimum;

[0045] S64: Obtain the point cloud p as shown in the following formula.j and p i Center of mass:

[0046] (9);

[0047] In the above formula, , Representing point cloud set p respectively j Hedianyunji p i The centroid; N represents the number of points in the point cloud cluster;

[0048] The following formulas decentralize point clouds respectively:

[0049] (10);

[0050] In the above formula, These are the decentralized points from the source point cloud and the target point cloud, respectively.

[0051] The covariance matrix H is constructed from the decentralized point sets of the source and target point clouds as follows:

[0052] (11);

[0053] The covariance matrix H is decomposed using SVD, and the decomposition result is as follows:

[0054] (12);

[0055] In the above formula, U represents the matrix representing the left singular vector of matrix H, and V represents the matrix representing the right singular vector of H;

[0056] S65: Calculate the rotation matrix R=VU T Translation vector ;

[0057] The calculated rotation matrix R and translation vector t are applied to the source point cloud as follows to obtain the source point cloud after pose update:

[0058] (13);

[0059] In the above formula, Let P be the source point cloud after multiple pose updates, and let P be the source point cloud after the last pose update.

[0060] S66: Determine whether the iteration should stop by calculating the mean square error of the two point clouds after the transformation, where the mean square error E is as follows:

[0061] (14);

[0062] S67: If the difference change is less than the preset threshold or the maximum number of iterations is reached, the iteration is terminated; otherwise, return to step S62 to continue the operation until the convergence condition is met; output the precise rotation axis parameter par.

[0063] 3. Beneficial effects:

[0064] (1) The present invention discloses a line structured light gear rotation scanning system, which differs from traditional CMM systems. It includes a central turntable and multiple line structured light sensors fixed on adjustable supports capable of precise height and attitude adjustment. In this system, a unified representation of the spatial relationships of each component can be achieved by establishing corresponding camera coordinate systems and base coordinate systems for multiple structured light sensors. Simultaneously, by determining the spatial transformation relationship between the camera coordinate system and the base coordinate system, the system determines the spatial position and direction of the rotation axis in the base coordinate system, thereby improving calibration accuracy. This calibration accuracy ensures the accuracy of point cloud rotation stitching, providing essential support for subsequent point cloud reconstruction.

[0065] (2) The present invention discloses a method for calibrating the rotation axis of a line structured light gear rotation scanning system. This method uses an articulated arm measuring machine and point cloud processing software to initially calibrate the transformation matrix. The articulated arm measuring machine scans and acquires a high-precision model of the workpiece, placing it at the origin horizontal position. The point cloud processing software matches the first frame of point cloud lines to the position of the model corresponding to the incident structural light rays on the workpiece. Based on the global movement of the point cloud lines, the initial transformation matrix is ​​obtained, achieving rapid initial calibration of the rotation axis. The calibration scheme, combined with the articulated arm measuring machine, is simple to operate, has a fast acquisition speed, high efficiency, and stable data point quality. Compared to the commonly used method of acquiring trajectory points on different planes and fitting circles on multiple planes, the selected rotation axis calibration method has a simplified and efficient calculation process.

[0066] (3) The present invention discloses a method for calibrating the rotating shaft of a line structure light gear rotation scanning system. Multiple line structure light sensors scan around a central fixed turntable. The central reference point is fixed, so there is no need to analyze and determine the reference point. At the same time, multiple sensors can quickly acquire the complete point cloud of the object to be measured. Attached Figure Description

[0067] Figure 1 This is an overall schematic diagram of the line structure optical gear rotary scanning system involved in this invention;

[0068] Figure 2 This is an external view of the standard part used in the specific embodiment;

[0069] Figure 3 This is a schematic diagram of the camera coordinate system and base coordinate system established in this method;

[0070] Figure 4This is a schematic diagram of the axial space transformation in steps S52 and S53 of the present invention;

[0071] Figure 5 This is a schematic diagram of the geometric transformation around an arbitrary axis in space in steps S52 and S53 of this method;

[0072] Figure 6 To verify the deviation diagram of the point cloud rotation splicing of standard parts before and after the optimization of the rotation axis in the example;

[0073] Figure 7 This is a flowchart of the rotation axis calibration method for the line structured light gear rotation scanning system of the present invention.

[0074] Figure reference numerals: 1. Linear structured light sensor; 2. Central turntable; 3. Workpiece mounting point; 4. Z-axis optical adjustment frame mechanism; 5. Precision angular displacement stage; 6. X-axis optical adjustment frame mechanism; 7. Y-axis optical adjustment frame mechanism. Detailed Implementation

[0075] The present invention will now be described in detail with reference to the accompanying drawings.

[0076] As attached Figure 1 To be continued Figure 6 As shown, a line structured light gear-like rotary scanning system is characterized by comprising multiple structured light sensors, a multi-degree-of-freedom optical adjustment frame, a precision angular displacement stage, an articulated arm measuring machine, and a central turntable. The multiple structured light sensors are mounted on the multi-degree-of-freedom optical adjustment frame and the angular displacement stage to acquire point cloud information of the workpiece from different perspectives. The multi-degree-of-freedom optical adjustment frame adjusts the structured light sensors to their positions on the XYZ axes, and the angular displacement stage fine-tunes the angle of the light sensors to obtain the optimal scanning pose. The laser scanning probe of the articulated arm measuring machine acquires the STL model of the workpiece, which is used to calibrate the accuracy of the point cloud acquired by the system. The central turntable is a disc-type turntable, using its own central axis as the motion reference, and achieves fixed-axis rotation around the central turntable axis through a drive mechanism. The workpiece is fixed to a preset positioning area on the central turntable surface and rotates synchronously with the rotation of the central turntable, thereby enabling the line structured light sensors to continuously scan and acquire data on the surface contour within the corresponding perspective during the workpiece's movement.

[0077] Furthermore, the end effector of the articulated arm measuring machine integrates a standardized connection structure, which has a modular interface function and can realize detachable fixation and precise positioning of the scanning probe.

[0078] As attached Figure 1The diagram shows a schematic of the hardware system of a line structured light gear-like rotary scanning system. The diagram uses four structured light sensors as an example; the number of sensors is adjusted according to the specific shape and size of the workpiece under test, ensuring that the viewing angle of all sensors covers the workpiece. The articulated arm measuring machine is not shown in the diagram. As shown, the system includes a multi-degree-of-freedom optical adjustment frame for adjusting the position of the structured light sensors, consisting of an X-axis optical adjustment frame mechanism 6, a Y-axis optical adjustment frame mechanism 7, and a Z-axis translation mechanism 4; an angular displacement stage for fine-tuning the angle of the structured light sensors; a central turntable for fixing the workpiece under test; and an articulated arm measuring machine for precise positioning. The adjustment frame and angular displacement stage improve the quality of a single image and the measurable area; the turntable reduces flipping and manual intervention; and the articulated arm ensures reliable calibration and controllable error. The appearance of the standard calibration component is shown in the attached figure. Figure 2 As shown, it includes various shapes, capable of encompassing the three-dimensional form of gears. Its appearance as a standard calibration part is shown in the attached figure. Figure 2 As shown, it includes various shapes and can cover the three-dimensional form of gears.

[0079] As attached Figure 7 As shown, a method for calibrating the rotating shaft of a line structured light gear-type rotary scanning system includes the following steps:

[0080] Step 1: Construct a line structured light gear-like rotary scanning system; place the standard part to be inspected on the central turntable; adjust the positions of multiple structured light sensors so that the sum of their viewing angles covers the entire outline of the object being inspected; install a laser scanning probe at the end of the articulated arm; as shown in the attached diagram. Figure 1 As shown;

[0081] Step 1: Set up a line structured light gear rotary scanning system; place the standard part to be inspected on the central turntable; adjust the positions of multiple structured light sensors so that the sum of the viewing angles of the multiple structured light sensors can cover the entire outline of the object to be measured; install a laser scanning probe at the end of the articulated arm measurement.

[0082] Step 2: Determine the base coordinate system of the line structured light gear rotation scanning system; take the intersection of the central rotation axis of the central turntable and the table surface as the origin of the base coordinate system, the Z-axis direction is the same as the rotation axis of the central turntable, and the plane formed by the X and Y axes is the plane of the central turntable table surface, which does not rotate with the rotation of the turntable.

[0083] Step 3: Determine the corresponding camera coordinate system for each structured light sensor: Select the camera optical center or the imaging center of the image plane of the sensor as the origin of the coordinate system. Its X and Z axes together constitute the light plane of the line structured light sensor. The Z-axis direction is parallel to and opposite to the line structured light projection direction, and the Y-axis is perpendicular to the light plane according to the right-hand rule.

[0084] The coordinate system constructed in this invention is detailed in the appendix. Figure 3 As shown, Figure 3 The O on the left represents the base coordinate system of the central turntable where the workpiece is placed; the L on the right represents the camera coordinate system of one of the structured light sensors. The base coordinate system serves as the global reference coordinate system for the rotating scanning system. This coordinate system is rigidly fixed to the base and does not change with the rotation of the turntable or the movement of other components. It is the original reference coordinate system used in system calibration and 3D reconstruction. The camera coordinate system, as the body coordinate system of the line structured light sensor, describes the spatial position of each point in the camera imaging plane and the laser light plane. This coordinate system directly relates the geometric mapping between the camera imaging coordinates and the spatial points of the workpiece, and is the original reference coordinate system used in system calibration and 3D reconstruction.

[0085] Step 4: Control the articulated arm measuring machine to scan the outer surface of the standard part to obtain its three-dimensional geometric model and output it in STL format to obtain a triangular mesh STL model; based on the triangular mesh STL model, use area weighting and point spacing constraints to generate a three-dimensional point cloud set that meets the preset density and distribution characteristics; this three-dimensional point cloud set is the calibrated target point cloud;

[0086] Step 5: Analyze all structured light sensors to obtain the initial transformation matrix T_C^B for transforming the camera coordinate system of the structured light sensors to the base coordinate system; preset the unit rotation angle and control the turntable to drive the standard part to rotate synchronously for one revolution. The structured light sensors at different positions collect the point cloud corresponding to each unit rotation angle. All the point clouds collected by the structured light sensors at the same position during this process constitute the complete elongated point cloud of the standard part's outline; based on the transformation matrix M between the base coordinate system and the camera coordinate system, each elongated point cloud is rotated and stitched together to obtain the rotated scan point cloud of the outline of the part under test.

[0087] Step 6: Based on ICP point cloud registration, define the rotated scan point cloud as the source point cloud, iteratively optimize the relative position and orientation between the target point cloud and the source point cloud, so that the distance between the corresponding point clouds gradually decreases, thereby minimizing the error and obtaining the rotation axis parameters of the current state.

[0088] Furthermore, the standard part is a geometric model of a polyhedron, and its shape is provided with the geometric shapes of planes, trapezoids, and holes.

[0089] Furthermore, in step four, the triangular mesh STL model is converted from a mesh to a point cloud using reverse modeling software.

[0090] Furthermore, step five specifically includes:

[0091] S51: Taking one of the structured light sensors as an example, the first frame point cloud obtained through reverse processing software is matched to the model position corresponding to the workpiece position where the incident light is located, thus obtaining the global translation matrix of the first frame point cloud, i.e., the initial transformation matrix. ;

[0092] As attached Figure 3 , 4 As shown in Figure 5, Figure 4 The coordinate system on the left side represents the base coordinate system O. O The coordinate system in which it is located is the camera coordinate system on the right; Figure 3 This reflects the field-of-view relationship of the point cloud of standard parts acquired by the camera; Figure 4 This indicates the mapping relationship between the axis in the base coordinate system and the camera coordinate system; Figure 5 This reflects the angular relationship of the camera's axis rotating around the XYZ axes in the camera coordinate system.

[0093] S52: Take the unit vector on the Z-axis of the base coordinate system. From the initial transformation matrix The vector n is obtained by inverse transformation to the camera coordinate system; the starting point of vector n is translated to the origin of the camera coordinate system, and the translation parameter is... Then the translation transformation matrix T is as follows:

[0094] (1);

[0095] In the above formula, t x t y t z These represent the translation parameters on the x, y, and z coordinate axes respectively during this process;

[0096] S53: Rotate sequentially around the X, Y, and Z axes of its own coordinate system from the initial state;

[0097] When rotating around the X-axis by an angle When, its corresponding rotation matrix R x ( ) is represented as:

[0098] (2);

[0099] When rotating around the Y-axis by an angle β, since counterclockwise rotation is defined as positive in a right-handed coordinate system, the corresponding rotation angle for clockwise rotation should be negative, and the corresponding rotation matrix R... y (-β) is as follows:

[0100] (3);

[0101] When rotating around the Z-axis by an angle γ, the corresponding rotation matrix R z(γ) is represented as:

[0102] (4);

[0103] Combining equations (2), (3), and (4), we obtain the rotation transformation matrix R. xyz As shown in the following formula:

[0104] (5);

[0105] In the above formula, β is the angle between the direction vector n of the current frame point cloud and the X-axis in the XOZ plane; β is the angle between the direction vector n of the current frame point cloud and the Y-axis in the YOZ plane; γ represents the rotation angle of the corresponding axis cylinder of the current frame point cloud around the Z-axis;

[0106] S54: Equation (5) is the transformation from the base coordinate system to the camera coordinate system. Therefore, the transformation matrix M can be obtained by its inverse process, that is:

[0107] (6).

[0108] As attached Figure 4 , 5 As shown, since the central turntable of the system is a shaft column that can fix the object to be measured, the Z-axis can be regarded as its central axis. Therefore, the unit coordinate point on the Z-axis in the base coordinate system is taken. and Calculate the initial axis located on the Z-axis. The transformation of the initial axis in different spaces is shown in the attached figure. Figure 4 As shown. Rotational geometric transformations about arbitrary axes in space are as follows: Figure 5 As shown.

[0109] S55: As in steps S51-54, until the transformation matrix M corresponding to all structured light sensors is obtained, and the transformation matrix M is used to realize the rotation and stitching of all strip-shaped point clouds.

[0110] Furthermore, step six involves designing and implementing a point cloud registration algorithm based on the ICP algorithm, and employing a grid grouping strategy and iterative optimization method for parameter solving and accelerated computation; the specific process includes:

[0111] S61: The preset rotation axis includes 11 parameters as shown in the following formula, and its parameter vector expression par is as follows:

[0112] (7);

[0113] In the above formula, r x， r y， r z These represent the rotation parameters for the corresponding coordinate axes; p x p y pz These are the coordinates of the starting point of the rotation axis corresponding to the coordinate axis; these 11 parameters can fully reflect the transformation matrix M;

[0114] S62: Perform denoising and outlier removal preprocessing on both the target point cloud and the source point cloud; perform downsampling on the preprocessed point cloud; and then use the processed target point cloud... Representation, source cloud usage express;

[0115] S63: Divide the target point cloud space into multiple small cells, each cell containing a local subset of the target point set. The initial grid step size S is:

[0116] (8);

[0117] In the above formula, h is a parameter used to scale the point cloud, and its size is set according to the total number of points; and These represent the maximum and minimum values ​​of the target point cloud along its x, y, and z axes, respectively; a point cloud set is taken from the target point cloud set. Find the corresponding point cloud set in the source point cloud set. , making Minimum;

[0118] S64: Obtain the point cloud p as shown in the following formula. j and p i Center of mass:

[0119] (9);

[0120] In the above formula, , Representing point cloud set p respectively j Hedianyunji p i The centroid; N represents the number of points in the point cloud cluster;

[0121] The following formulas decentralize point clouds respectively:

[0122] (10);

[0123] In the above formula, These are the decentralized points from the source point cloud and the target point cloud, respectively.

[0124] The covariance matrix H is constructed from the decentralized point sets of the source and target point clouds as follows:

[0125] (11);

[0126] The covariance matrix H is decomposed using SVD, and the decomposition result is as follows:

[0127] (12);

[0128] In the above formula, U represents the matrix representing the left singular vector of matrix H, and V represents the matrix representing the right singular vector of H;

[0129] S65: Calculate the rotation matrix R=VU T Translation vector ;

[0130] The calculated rotation matrix R and translation vector t are applied to the source point cloud as follows to obtain the source point cloud after pose update:

[0131] (13);

[0132] In the above formula, Let P be the source point cloud after multiple pose updates, and let P be the source point cloud after the last pose update.

[0133] S66: Determine whether the iteration should stop by calculating the mean square error of the two point clouds after the transformation, where the mean square error E is as follows:

[0134] (14);

[0135] S67: If the difference change is less than the preset threshold or the maximum number of iterations is reached, the iteration is terminated; otherwise, return to step S62 to continue the operation until the convergence condition is met; output the precise rotation axis parameter par.

[0136] Verification example:

[0137] To verify the calibration effect of this scheme, the following example is used: Figure 2 The standard part shown is used as a calibration reference. It has a good three-dimensional morphology. The line structured light sensor collects the point cloud of the standard part once every 1° rotation of the turntable. Figure 6 This image shows the color deviation of the point cloud before and after optimization using this method. The image shows that the accuracy range before optimization was 0.08-0.431, while the accuracy range after optimization was 0.01-0.048. The image also shows that the large color range before optimization indicates a large error range and poor overall fit. After optimization, the colors are more concentrated in neutral colors (such as green / light yellow), indicating a decrease in both the average and maximum errors, better overall fit, higher accuracy, and a significant calibration effect.

[0138] Although the present invention has been disclosed above with reference to preferred embodiments, these are not intended to limit the invention. Any person skilled in the art can make various changes or modifications without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention should be defined by the scope of the claims of this application.

Claims

1. A line structured light gear-like rotary scanning system, characterized in that: The system includes multiple structured light sensors, a multi-degree-of-freedom optical adjustment frame, a precision angular displacement stage, an articulated arm measuring machine, and a central turntable. The multiple structured light sensors are mounted on the multi-degree-of-freedom optical adjustment frame and the angular displacement stage to acquire point cloud information of the workpiece from different perspectives. The multi-degree-of-freedom optical adjustment frame adjusts the structured light sensors to position them on the XYZ axes, and the angular displacement stage fine-tunes the angle of the light sensors to obtain the optimal scanning pose. The laser scanning probe of the articulated arm measuring machine acquires the STL model of the workpiece, which is used to calibrate the accuracy of the point cloud acquired by the system. The central turntable is a disc-type turntable, using its own central axis as the motion reference, and a drive mechanism achieves fixed-axis rotational motion around the central turntable axis. The workpiece is fixed to a preset positioning area on the central turntable and rotates synchronously with the rotation of the central turntable, thereby enabling the line structured light sensor to continuously scan and acquire data on the surface contour within the corresponding viewpoint during the workpiece's movement.

2. The line structured light gear-like rotary scanning system according to claim 1, characterized in that: The end effector of the articulated arm measuring machine integrates a standardized connection structure with a modular interface function, which enables detachable fixing and precise positioning of the scanning probe.

3. A method for calibrating the rotating shaft of a line-structured optical gear-like rotary scanning system as described in claim 1 or 2, characterized in that: Includes the following steps: Step 1: Set up a line structured light gear rotary scanning system; place the standard part to be inspected on the central turntable; adjust the positions of multiple structured light sensors so that the sum of the viewing angles of the multiple structured light sensors can cover the entire outline of the object to be measured; install a laser scanning probe at the end of the articulated arm measurement; Step 2: Determine the base coordinate system of the line structure light gear rotary scanning system; take the intersection of the central rotation axis of the central turntable and the table surface as the origin of the base coordinate system, the Z-axis direction is the same as the rotation axis of the central turntable, and the plane formed by the X and Y axes is the plane of the central turntable table surface, which does not rotate with the rotation of the turntable. Step 3: Determine the corresponding camera coordinate system for each structured light sensor: Select the camera optical center or the imaging center of the image plane of the sensor as the origin of the coordinate system. Its X and Z axes together constitute the light plane of the line structured light sensor. The Z-axis direction is parallel to and opposite to the line structured light projection direction, and the Y-axis is perpendicular to the light plane according to the right-hand rule. Step 4: Control the articulated arm measuring machine to scan the outer surface of the standard part to obtain its three-dimensional geometric model and output it in STL format to obtain a triangular mesh STL model; based on the triangular mesh STL model, use area weighting and point spacing constraints to generate a three-dimensional point cloud set that meets the preset density and distribution characteristics; this three-dimensional point cloud set is the calibrated target point cloud; Step 5: Analyze all structured light sensors to obtain the initial transformation matrix from the camera coordinate system to the base coordinate system. ; A preset unit rotation angle is used to control the turntable to drive the standard part to rotate synchronously for one revolution. The structure light sensors at different positions collect the point cloud corresponding to each unit rotation angle. All the point clouds acquired by the structure light sensor at the same position during this process constitute a complete long strip point cloud of the standard part's outline. Based on the transformation matrix M between the base coordinate system and the camera coordinate system, each long strip point cloud is rotated and stitched together to obtain the rotated scan point cloud of the outline of the part under test. Step 6: Based on ICP point cloud registration, define the rotated scan point cloud as the source point cloud, iteratively optimize the relative position and orientation between the target point cloud and the source point cloud, so that the distance between the corresponding point clouds gradually decreases, thereby minimizing the error and obtaining the rotation axis parameters of the current state.

4. The method for calibrating the rotating shaft of a line structured light gear-like rotary scanning system according to claim 3, characterized in that: The standard part is a geometric model of a polyhedron, and its shape is set with the geometric shapes of planes, trapezoids, and holes.

5. The method for calibrating the rotating shaft of a linear three-dimensional gear optical scanning system according to claim 3, characterized in that: In step four, the triangular mesh STL model is converted from a mesh to a point cloud using reverse modeling software.

6. The method for calibrating the rotating shaft of a line structured light gear-like rotary scanning system according to claim 3, characterized in that: Step five specifically includes: S51: Taking one of the structured light sensors as an example, the first frame point cloud obtained through reverse processing software is matched to the model position corresponding to the workpiece position where the incident light is located, thus obtaining the global translation matrix of the first frame point cloud, i.e., the initial transformation matrix. ; S52: Take the unit vector on the Z-axis of the base coordinate system. From the initial transformation matrix The vector n is obtained by inverse transformation to the camera coordinate system; the starting point of vector n is translated to the origin of the camera coordinate system, and the translation parameter is... Then the translation transformation matrix T is as follows: （1）； In the above formula, t x , t y , t z respectively represent the translation parameters in the x, y, z coordinate axes of the process; S53: Rotate sequentially around the X, Y, and Z axes of its own coordinate system from the initial state; When rotating around the X-axis by an angle When, its corresponding rotation matrix R x ( ) is represented as: （2）； When rotating around the Y-axis by an angle β, since counterclockwise rotation is defined as positive in a right-handed coordinate system, the corresponding rotation angle for clockwise rotation should be negative, and the corresponding rotation matrix R... y (-β) is as follows: （3）； When rotating around the Z-axis by an angle γ, the corresponding rotation matrix R z (γ) is represented as: （4）； Combining equations (2), (3), and (4), we obtain the rotation transformation matrix R. xyz As shown in the following formula: （5）； In the above formula, β is the angle between the direction vector n of the current frame point cloud and the X-axis in the XOZ plane; β is the angle between the direction vector n of the current frame point cloud and the Y-axis in the YOZ plane; γ represents the rotation angle of the corresponding axis cylinder of the current frame point cloud around the Z-axis; S54: Equation (5) is the transformation from the base coordinate system to the camera coordinate system. Therefore, the transformation matrix M can be obtained by its inverse process, that is: （6）； S55: As in steps S51-54, until the transformation matrix M corresponding to all structured light sensors is obtained, and the transformation matrix M is used to realize the rotation and stitching of all strip-shaped point clouds.

7. The method for calibrating the rotating shaft of a linear three-dimensional gear optical scanning system according to claim 1, characterized in that: In step six, a point cloud registration algorithm based on the ICP algorithm is designed and implemented, and a grid grouping strategy and iterative optimization method are used to solve parameters and accelerate calculation. The process specifically includes: S61: The preset rotation axis includes 11 parameters as shown in the following formula, and its parameter vector expression par is as follows: （7）； In the above formula, r x， r y， r z These represent the rotation parameters for the corresponding coordinate axes; p x p y p z These are the coordinates of the starting point of the rotation axis corresponding to the coordinate axis; these 11 parameters can fully reflect the transformation matrix M; S62: Perform denoising and outlier removal preprocessing on both the target point cloud and the source point cloud; perform downsampling on the preprocessed point cloud; and then use the processed target point cloud... Representation, source cloud usage express; S63: Divide the target point cloud space into multiple small cells, each cell containing a local subset of the target point set. The initial grid step size S is: （8）； In the above formula, h is a parameter used to scale the point cloud, and its size is set according to the total number of points; and These represent the maximum and minimum values ​​of the target point cloud along its x, y, and z axes, respectively; a point cloud set is taken from the target point cloud set. Find the corresponding point cloud set in the source point cloud set. This makes the target point cloud p j With source point cloud p i Relative position between Minimum; S64: Obtain the point cloud p as shown in the following formula. j and p i Center of mass: （9）； In the above formula, , Representing point cloud set p respectively j Hedianyunji p i The centroid; N represents the number of points in the point cloud cluster; The following formulas decentralize point clouds respectively: （10）； In the above formula, These are the decentralized points from the source point cloud and the target point cloud, respectively. The covariance matrix H is constructed from the decentralized point sets of the source and target point clouds as follows: （11）； The covariance matrix H is decomposed using SVD, and the decomposition result is as follows: （12）； In the above formula, U represents the matrix representing the left singular vector of matrix H, and V represents the matrix representing the right singular vector of H; S65: Calculate the rotation matrix R=VU T Translation vector ; The calculated rotation matrix R and translation vector t are applied to the source point cloud as follows to obtain the source point cloud after pose update: （13）; In the above formula, Let P be the source point cloud after multiple pose updates, and let P be the source point cloud after the last pose update. S66: Determine whether the iteration should stop by calculating the mean square error of the two point clouds after the transformation, where the mean square error E is as follows: （14）； S67: If the difference change is less than the preset threshold or the maximum number of iterations is reached, the iteration is terminated; otherwise, return to step S62 to continue the operation until the convergence condition is met; output the precise rotation axis parameter par.

Images