Structured light splicing method and system under three-coordinates

Abstract

The present invention discloses a method and system for splicing structured light under three-coordinates, which relates to the field of image processing. The system includes: an acquisition module for projecting a composite coding pattern through a structured light projection device, cooperating with an industrial camera to acquire deformed stripe images on the workpiece surface from preset multiple angles, and synchronously acquiring point cloud data under the structured light coordinate system, covering the full surface features of the workpiece; the present invention adopts a composite coding pattern combining Gray code and phase-shifted stripes, combined with adaptive exposure control and curvature-based path planning, to achieve high-precision acquisition of complex surface details, with phase calculation accuracy reaching sub-pixel level, and the scanning step size dynamically adjusted according to the surface curvature, effectively improving the acquisition accuracy of complex areas.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method and system for splicing structured light in three coordinates. Background Art

[0002] Structured light 3D measurement projects a composite coded pattern, and an industrial camera captures images of the workpiece's deformed stripes from multiple angles to generate point cloud data. Combined with a three-dimensional coordinate measuring machine (CMM) moving the probe to cover the entire surface, the system then generates a complete 3D point cloud model after denoising, coordinate calibration, and coarse and fine registration, enabling high-precision 3D reconstruction of the workpiece.

[0003] The invention patent application with application number 202011490117.4 discloses a multi-camera large field of view surface shape measurement calibration method. The application aims to solve the problem that "current structured light calibration methods are mostly calibrated for monocular cameras. For the same monocular structured light measurement system, the larger the measured workpiece area and volume, the lower the measurement accuracy. When reconstructing the three-dimensional morphology of ultra-large workpieces, the single-camera system is bound to be limited."

[0004] However, existing structured light 3D measurement methods suffer from problems such as insufficient precision in multi-view point cloud stitching, incomplete acquisition of complex surface details, and excessive noise caused by ambient light interference. Furthermore, traditional methods have limited compensation for equipment motion errors, making it difficult to achieve high-precision 3D reconstruction of the entire workpiece surface. Optimization of point cloud acquisition, calibration, and registration technologies is needed to improve measurement results.

[0005] Therefore, a three-coordinate structured light splicing method and system are proposed. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the prior art, the present invention provides a method and system for splicing structured light in three coordinates, which can effectively solve the problems of the prior art.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] The present invention discloses a three-coordinate structured light splicing system, comprising:

[0009] The acquisition module is used to project a composite coding pattern through a structured light projection device, cooperate with the industrial camera to collect the deformed stripe image of the workpiece surface from preset multiple angles, and synchronously obtain the point cloud data in the structured light coordinate system to cover the full surface features of the workpiece; the motion control module is used to drive the X, Y, and Z axis servo motors and linear guides of the three-dimensional coordinate measuring machine, and move the probe of the integrated structured light system according to the preset trajectory to ensure full coverage of the measurement area, so that the overlap rate of adjacent scanning areas is 20%-30% of the average area of ​​adjacent scanning areas, and the motion positioning accuracy is controlled within ±0.01mm; the pre-processing module is used to denoise, smooth and resample the original point cloud data to eliminate interference from ambient light and surface reflections. The noise points generated are used to optimize the uniformity of point cloud density; the calibration conversion module is used to establish the initial conversion relationship between the structured light coordinate system and the world coordinate system of the three-dimensional coordinate measuring machine through high-precision calibration tools such as standard balls, and dynamically update the calibration parameters during the measurement process to compensate for the motion error of the equipment; the stitching module is used to extract the corner points and geometric features of the planes in the overlapping areas of adjacent point clouds and match them, use the RANSAC algorithm to complete the coarse alignment to provide an initial pose estimation, and apply the iterative closest point algorithm and its improved algorithm for fine alignment; the output module is used to convert the aligned point cloud data to the world coordinate system, apply the averaging method or surface reconstruction algorithm to fuse the overlapping areas, and generate a complete three-dimensional point cloud model.

[0010] Furthermore, the composite coding pattern is a combined coding pattern of Gray code and phase-shift stripes;

[0011] Gray code is used for global phase unwrapping, phase shift stripes are used for sub-pixel phase calculation, and the phase main value calculation formula is:

[0012] ;

[0013] Where: are the light intensity values ​​of the four-step phase-shifted fringe images respectively.

[0014] Furthermore, the acquisition frame rate of the industrial camera is synchronized with the pattern projection rate of the structured light projection device, and the acquisition exposure time of a single image is adaptively adjusted according to the ambient light intensity. The adjustment formula is:

[0015] ;

[0016] Where: Acquire the exposure time for the adjusted single image; is the exposure gain coefficient of the camera; is the ambient light intensity detection value; is the exposure bias coefficient of the camera; is the image grayscale saturation threshold;

[0017] Among them, the image grayscale saturation threshold Preset for system-side users.

[0018] Furthermore, the preset trajectory of the motion control module adopts a curvature-based path planning algorithm to automatically adjust the scanning step size according to the curvature change of the workpiece surface. The curvature calculation formula is:

[0019] ;

[0020] Where: It is the second-order derivative of the function expression of the workpiece surface curve, which represents the rate of change of the curvature of the point on the curve; It is the first derivative of the function expression of the workpiece surface curve, which represents the slope of the tangent line at the point on the curve;

[0021] Among them, when the curvature When the value is greater than the preset threshold, the scanning step is automatically reduced to half of the standard step.

[0022] Furthermore, the preset trajectory of the motion control module adopts a curvature-based path planning algorithm to automatically adjust the scanning step size according to the curvature change of the workpiece surface. The curvature calculation formula is:

[0023] ;

[0024] Where: It is the second-order derivative of the function expression of the workpiece surface curve, which represents the rate of change of the curvature of the point on the curve; It is the first derivative of the function expression of the workpiece surface curve, which represents the slope of the tangent line at the point on the curve;

[0025] Among them, when the curvature When the value is greater than the preset threshold, the scanning step is automatically reduced to half of the standard step.

[0026] Furthermore, the calibration conversion module adopts a multi-station calibration method based on a standard sphere, and fits the center coordinates of the standard sphere in the structured light coordinate system by the least square method. The coordinates of the sphere center in the world coordinate system , establish the homogeneous transformation matrix ;

[0027] ;

[0028] Where: Homogeneous transformation matrix It contains the rotation matrix R and the translation vector t, and compensates the motion error of the coordinate measuring machine through iterative optimization.

[0029] Furthermore, in the coarse registration stage of the stitching module, a feature matching algorithm based on FAST corner detection and FREAK feature descriptor is used to eliminate mismatched point pairs through Lowe's ratio test, and the matching efficiency is improved by more than 30% compared with the traditional SIFT algorithm.

[0030] In the fine registration stage of the stitching module, an improved ICP algorithm is used to control the registration root mean square error within 0.005 mm by introducing a point cloud normal vector constraint term.

[0031] Furthermore, the surface reconstruction algorithm of the output module adopts surface fitting based on the moving least squares method, and by constructing a local quadratic polynomial to approximate the point cloud surface, the generated triangular mesh model satisfies the Euler formula:

[0032] ;

[0033] Where: is the number of vertices; is the number of sides; is the number of faces; is the grid genus;

[0034] Among them, the maximum ratio of the side length to the height of the grid is less than 8, and the minimum internal angle is greater than 20 degrees.

[0035] Furthermore, the acquisition module is interactively connected to the motion control module through a wireless network, the motion control module is interactively connected to the preprocessing module, i.e., the calibration conversion module, through a wireless network, the calibration conversion module is interactively connected to the splicing module through a wireless network, and the splicing module is interactively connected to the output module through a wireless network.

[0036] On the other hand, the splicing method of structured light under three-coordinates includes:

[0037] The structured light projection device is started to project a composite coded pattern, and the industrial camera synchronously collects the deformed stripe image of the workpiece surface. A single acquisition covers a single measurement area, and the original image data of each measurement area and the point cloud data in the structured light coordinate system are output; according to the size and shape of the workpiece, the probe is driven to move according to the preset path to complete the scanning of each area and form a multi-view point cloud data set; the original point cloud is denoised, smoothed and resampled to output the processed point cloud data of each area; the structured light coordinate system and the world coordinate system are mapped through the coordinate system calibration and conversion module, so that the coordinate conversion matrix of the point cloud of each area is converted from the structured light coordinate system to the world coordinate system; by extracting and matching the geometric features of adjacent areas, the random sampling consensus algorithm is applied to complete the coarse alignment, and then the iterative nearest point algorithm is used to optimize the parameters to complete the fine alignment; the point cloud after fine alignment is unified to the world coordinate system, and the overlapping areas are fused to generate a three-dimensional point cloud model, which is further converted into a triangular mesh model.

[0038] Compared with the prior art, the technical solution provided by the present invention has the following beneficial effects:

[0039] The present invention provides a method and system for splicing structured light in three-dimensional coordinates. During execution, this method and system use a composite coding pattern combining Gray code and phase-shifted stripes, combined with adaptive exposure control and curvature-based path planning, to achieve high-precision acquisition of complex surface details. Phase calculation accuracy reaches sub-pixel level, and the scanning step size is dynamically adjusted according to the surface curvature, effectively improving acquisition accuracy in complex areas.

[0040] A bilateral filtering algorithm was introduced in the preprocessing stage, using dual-weighted adaptive denoising based on spatial proximity and grayscale similarity to effectively eliminate noise while preserving the edge features of the point cloud. For coordinate calibration, a multi-station calibration model based on a standard sphere, combined with the least squares method and iterative optimization, dynamically compensated for device motion errors and established a high-precision mapping relationship between the structured light coordinate system and the world coordinate system.

[0041] During the stitching process, the improved coarse and fine registration algorithms significantly improved the registration efficiency and accuracy. The matching algorithm based on FAST corner points and FREAK features was 30% more efficient than the traditional method. The ICP fine registration with normal vector constraints controlled the root mean square error within 0.005mm. Finally, the moving least squares surface reconstruction with topological constraints was used to generate a triangular mesh model that satisfied the Euler formula, ensuring the geometric accuracy and topological correctness of the three-dimensional model. The overall automation level, measurement range and adaptability to complex scenes were all optimized compared to existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0043] Figure 1 The schematic diagram of the structure of the splicing system of structured light in three coordinates;

[0044] Figure 2 The figure is a flowchart of the structured light stitching method under three coordinates. DETAILED DESCRIPTION

[0045] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0046] The present invention will be further described below with reference to the embodiments.

[0047] Example 1:

[0048] The three-coordinate structured light splicing system of this embodiment is as follows: Figure 1 As shown, including:

[0049] The acquisition module is used to project a composite coded pattern through a structured light projection device, cooperate with the industrial camera to collect deformed stripe images on the workpiece surface from preset multiple angles, and simultaneously obtain point cloud data in the structured light coordinate system to cover the full surface features of the workpiece;

[0050] The composite coding pattern is a combination coding pattern of Gray code and phase-shifted stripes;

[0051] Gray code is used for global phase unwrapping, phase shift stripes are used for sub-pixel phase calculation, and the phase main value calculation formula is:

[0052] ;

[0053] Where: are the light intensity values ​​of the four-step phase-shifted fringe images;

[0054] The main phase value is calculated by the four-step phase-shifted stripe light intensity value, and the sub-pixel phase solution of the combined coding pattern of Gray code and phase-shifted stripes is realized, thereby improving the decoding accuracy of structured light stripes.

[0055] The acquisition frame rate of the industrial camera is synchronized with the pattern projection rate of the structured light projection device, and the acquisition exposure time of a single image is adaptively adjusted according to the ambient light intensity. The adjustment formula is:

[0056] ;

[0057] Where: Acquire the exposure time for the adjusted single image; is the exposure gain coefficient of the camera; is the ambient light intensity detection value; is the exposure bias coefficient of the camera; is the image grayscale saturation threshold;

[0058] Among them, the image grayscale saturation threshold Preset for system end users;

[0059] The camera exposure time is adaptively adjusted according to the ambient light intensity to ensure that the grayscale value of the image collected by the industrial camera under different lighting conditions is stable within a reasonable range, avoiding the distortion of fringe information caused by overexposure or underexposure.

[0060] The motion control module is used to drive the X, Y, and Z axis servo motors and linear guides of the three-dimensional coordinate measuring machine, and move the probe of the integrated structured light system according to the preset trajectory to ensure full coverage of the measurement area, so that the overlap rate of adjacent scanning areas is 20%-30% of the average area of ​​adjacent scanning areas, and the motion positioning accuracy is controlled within ±0.01mm;

[0061] The preset trajectory of the motion control module adopts a curvature-based path planning algorithm to automatically adjust the scanning step size according to the curvature change of the workpiece surface. The curvature calculation formula is:

[0062] ;

[0063] Where: It is the second-order derivative of the function expression of the workpiece surface curve, which represents the rate of change of the curvature of the point on the curve; It is the first derivative of the function expression of the workpiece surface curve, which represents the slope of the tangent line at the point on the curve;

[0064] Among them, when the curvature When it is greater than the preset threshold, the scanning step size is automatically reduced to half of the standard step size;

[0065] The scanning step length is dynamically adjusted based on the curvature calculation formula, so that the probe can automatically reduce the step length in complex curved surface areas with large curvature of the workpiece, improve the point cloud acquisition density, and ensure the complete capture of detailed features.

[0066] The pre-processing module is used to denoise, smooth and resample the original point cloud data, eliminate noise points caused by ambient light and surface reflection interference, and optimize the uniformity of point cloud density;

[0067] The preprocessing module uses bilateral filtering algorithm for denoising, and the filtering formula is:

[0068] ;

[0069] Where: is the pixel after filtering Gray value of is the pixel point in the original point cloud data Gray value of is the filter window; is the spatial proximity weight; is the grayscale similarity weight;

[0070] The bilateral filtering formula is combined with spatial proximity and grayscale similarity weights to remove point cloud noise while retaining edge features, thus achieving edge-preserving denoising optimization of point cloud data.

[0071] The calibration conversion module is used to establish the initial conversion relationship between the structured light coordinate system and the world coordinate system of the three-dimensional coordinate measuring machine using high-precision calibration tools such as standard spheres, and dynamically update the calibration parameters during the measurement process to compensate for the equipment motion error;

[0072] The calibration conversion module adopts a multi-station calibration method based on a standard sphere, and fits the center coordinates of the standard sphere in the structured light coordinate system through the least squares method. The coordinates of the sphere center in the world coordinate system , establish the homogeneous transformation matrix ;

[0073] ;

[0074] Where: Homogeneous transformation matrix Contains the rotation matrix R and the translation vector t, which compensates the motion error of the coordinate measuring machine through iterative optimization;

[0075] The mapping relationship between the structured light coordinate system and the world coordinate system is established through the homogeneous transformation matrix. Combined with the multi-station calibration of the standard sphere and the least squares fitting method, the motion error of the three-dimensional coordinate measuring machine is compensated and the coordinate conversion accuracy is improved.

[0076] The stitching module is used to extract the corner points and geometric features of the plane in the overlapping area of ​​adjacent point clouds and match them. The RANSAC algorithm is used to perform coarse registration to provide an initial pose estimate, and the iterative closest point algorithm and its improved algorithm are used for fine registration.

[0077] In the coarse registration stage of the stitching module, a feature matching algorithm based on FAST corner detection and FREAK feature descriptors is used. Mismatched point pairs are eliminated through the Lowe's ratio test, improving matching efficiency by over 30% compared to the traditional SIFT algorithm.

[0078] In the fine registration stage of the stitching module, an improved ICP algorithm is used to control the registration root mean square error within 0.005mm by introducing the point cloud normal vector constraint term;

[0079] The output module is used to convert the registered point cloud data into the world coordinate system, fuse the overlapping areas using the averaging method or surface reconstruction algorithm, and generate a complete 3D point cloud model;

[0080] The surface reconstruction algorithm of the output module adopts surface fitting based on the moving least squares method. By constructing a local quadratic polynomial to approximate the point cloud surface, the generated triangular mesh model satisfies the Euler formula:

[0081] ;

[0082] Where: is the number of vertices; is the number of sides; is the number of faces; is the grid genus;

[0083] Among them, the maximum ratio of the side length to the height of the grid is less than 8, and the minimum internal angle is greater than 20 degrees;

[0084] The surface reconstruction algorithm based on the moving least squares method is combined with the Euler formula constraint to ensure that the vertices, edges, and faces of the generated triangular mesh model meet the topological relationship, avoid geometric topological errors in the model, and ensure the integrity and accuracy of the 3D model;

[0085] The acquisition module is interactively connected to the motion control module through a wireless network, the motion control module is interactively connected to the preprocessing module, namely the calibration conversion module, through a wireless network, the calibration conversion module is interactively connected to the splicing module through a wireless network, and the splicing module is interactively connected to the output module through a wireless network.

[0086] In this embodiment, the acquisition module projects a composite coding pattern through a structured light projection device, cooperates with the industrial camera to collect the deformed stripe image of the workpiece surface from preset multiple angles, and synchronously obtains the point cloud data in the structured light coordinate system to cover the full surface features of the workpiece. The motion control module is rear-mounted to drive the X, Y, and Z axis servo motors and linear guides of the three-dimensional coordinate measuring machine, and moves the probe of the integrated structured light system according to the preset trajectory to ensure full coverage of the measurement area, so that the overlap rate of adjacent scanning areas is 20%-30% of the average area of ​​the adjacent scanning areas, and the motion positioning accuracy is controlled within ±0.01mm. The original point cloud data is denoised, smoothed, and resampled through the pre-processing module to eliminate ambient light and surface Noise points generated by reflective interference are eliminated to optimize the uniformity of point cloud density. The calibration conversion module standard ball and other high-precision calibration tools are further used to establish the initial conversion relationship between the structured light coordinate system and the world coordinate system of the three-dimensional coordinate measuring machine. The calibration parameters are dynamically updated during the measurement process to compensate for the motion error of the equipment. The stitching module then extracts the corner points and geometric features of the plane in the overlapping areas of adjacent point clouds and matches them. The RANSAC algorithm is used to complete the coarse alignment to provide an initial pose estimation. The iterative closest point algorithm and its improved algorithm are used for fine alignment. Finally, the output module converts the aligned point cloud data to the world coordinate system, and the averaging method or surface reconstruction algorithm is used to fuse the overlapping areas to generate a complete three-dimensional point cloud model.

[0087] In the above embodiment, the system improves the acquisition accuracy through composite coding patterns, adaptive acquisition and curvature dynamic scanning, combines bilateral filtering denoising, multi-station calibration and improved registration algorithm to achieve 0.005mm level registration accuracy, and finally generates a topologically correct three-dimensional model. Compared with the existing technology, it has higher adaptability and accuracy in complex scenes.

[0088] Additional notes:

[0089] Phase calculation logic of composite coding pattern:

[0090] Gray code patterns achieve global phase unwrapping through the unique binary encoding of adjacent stripes, avoiding the period ambiguity inherent in traditional phase-shifting methods. Taking an 8-bit Gray code as an example, the phase period can be expanded to 256. Combined with the sub-pixel accuracy of four-step phase-shifting stripes, the final 3D coordinate error of the point cloud can be controlled within 0.02mm.

[0091] Trajectory planning strategy for motion control:

[0092] The preset trajectory adopts the "layered scanning + adaptive step size" mode: for flat areas, the scanning step size is set to 0.5mm; for curved surface areas with large curvature, the step size is automatically reduced to 0.25mm based on the curvature calculation results, ensuring that the point cloud density of complex surfaces reaches more than 80 points per square millimeter.

[0093] Noise suppression mechanism of the preprocessing module:

[0094] Bilateral filtering uses dual weighting of the spatial coordinates and normal vectors of the three-dimensional point cloud: the spatial proximity weight is calculated based on the Euclidean distance, and the grayscale similarity weight is based on the cosine value of the normal vector angle (for example, the weight is attenuated by 50% when the angle is greater than 30°), removing outliers while retaining sharp edge features.

[0095] also:

[0096] In the coarse registration stage, FAST corner detection combined with the FREAK feature descriptor reduces computational complexity by 60% compared to the SIFT algorithm, shortening the matching time from 25 seconds to 7 seconds for a 1 million point cloud scenario. The Lowe's ratio test (with a threshold set to 0.6) keeps the mismatch rate below 5%.

[0097] Fine registration stage: The improved ICP algorithm introduces point cloud normal vector constraints, reducing the number of iterations from the traditional 50 to 15, and the registration root mean square error (RMSE) is controlled within 0.005mm.

[0098] Topology optimization strategies for model output

[0099] When generating the triangular mesh, a quadratic error metric is used to simplify the mesh, reducing the number of mesh faces by 60% while retaining the features. At the same time, the Laplace smoothing algorithm is used to optimize the mesh quality, ensuring that the minimum dihedral angle is greater than 30° and the maximum dihedral angle is less than 150°, meeting the requirements of 3D printing and CAE analysis.

[0100] Example 2:

[0101] In terms of specific implementation, based on Example 1, this example refers to Figure 2 The structured light splicing system under three coordinates in Example 1 is further described in detail:

[0102] The stitching method of structured light under three coordinates includes:

[0103] Step 1: Start the structured light projection device to project the composite coded pattern. The industrial camera synchronously collects the deformation stripe image on the workpiece surface. A single acquisition covers a single measurement area, and the original image data of each measurement area and the point cloud data in the structured light coordinate system are output;

[0104] Step 2: Based on the size and shape of the workpiece, the probe is driven to move along a preset path to complete the scanning of each area and form a multi-view point cloud data set;

[0105] Step 3: De-noise, smooth, and resample the original point cloud to output the processed point cloud data of each area;

[0106] Step 4: Use the coordinate system calibration and conversion module to establish the mapping between the structured light coordinate system and the world coordinate system, so that the coordinate conversion matrix of each area point cloud is converted from the structured light coordinate system to the world coordinate system;

[0107] Step 5: By extracting and matching the geometric features of adjacent regions, the random sampling consensus algorithm is applied to complete the coarse registration, and then the iterative closest point algorithm is used to optimize the parameters to complete the fine registration;

[0108] Step 6: Unify the precisely registered point cloud into the world coordinate system, fuse the overlapping areas to generate a 3D point cloud model, and further convert the 3D point cloud model into a triangular mesh model.

[0109] In summary, during the execution of the methods and systems in the above embodiments, a composite coding pattern combining Gray code and phase-shift stripes is adopted, combined with adaptive exposure control and curvature-based path planning, to achieve high-precision acquisition of complex surface details. The phase calculation accuracy reaches the sub-pixel level, and the scanning step size is dynamically adjusted according to the surface curvature, effectively improving the acquisition accuracy of complex areas. A bilateral filtering algorithm is introduced in the preprocessing stage, and dual-weighted adaptive denoising is performed through spatial proximity and grayscale similarity, effectively eliminating noise while retaining the edge features of the point cloud. In terms of coordinate calibration, the multi-station calibration model based on the standard sphere is combined with the least squares method and iterative optimization to dynamically compensate for the motion error of the equipment and establish a high-precision mapping relationship between the structured light coordinate system and the world coordinate system. During the stitching process, the improved coarse and fine registration algorithms significantly improved the registration efficiency and accuracy. The matching algorithm based on FAST corner points and FREAK features is 30% more efficient than the traditional method. The ICP fine registration with normal vector constraints controls the root mean square error within 0.005mm. Finally, through the moving least squares method with topological constraints, a triangular mesh model that satisfies the Euler formula is generated, ensuring the geometric accuracy and topological correctness of the three-dimensional model.

[0110] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. The structured light splicing system under three-coordinates is characterized by: include: The acquisition module is used to project a composite coded pattern through a structured light projection device, cooperate with the industrial camera to collect deformed stripe images on the workpiece surface from preset multiple angles, and simultaneously obtain point cloud data in the structured light coordinate system to cover the full surface features of the workpiece; The motion control module is used to drive the X, Y, and Z axis servo motors and linear guides of the three-dimensional coordinate measuring machine, and move the probe of the integrated structured light system according to the preset trajectory to ensure full coverage of the measurement area, so that the overlap rate of adjacent scanning areas is 20%-30% of the average area of ​​adjacent scanning areas, and the motion positioning accuracy is controlled within ±0.01mm; The pre-processing module is used to denoise, smooth and resample the original point cloud data, eliminate noise points caused by ambient light and surface reflection interference, and optimize the uniformity of point cloud density; The calibration conversion module is used to establish the initial conversion relationship between the structured light coordinate system and the world coordinate system of the three-dimensional coordinate measuring machine using high-precision calibration tools such as standard spheres, and dynamically update the calibration parameters during the measurement process to compensate for the equipment motion error; The stitching module is used to extract the corner points and geometric features of the plane in the overlapping area of ​​adjacent point clouds and match them. The RANSAC algorithm is used to perform coarse registration to provide an initial pose estimate, and the iterative closest point algorithm and its improved algorithm are used for fine registration. The output module is used to convert the registered point cloud data into the world coordinate system, apply the averaging method or surface reconstruction algorithm to fuse the overlapping areas, and generate a complete 3D point cloud model.

2. The three-coordinate structured light splicing system according to claim 1, characterized in that: The composite coding pattern is a combined coding pattern of Gray code and phase-shift stripes; Gray code is used for global phase unwrapping, phase shift stripes are used for sub-pixel phase calculation, and the phase main value calculation formula is: Where: I1(x,y), I2(x,y), I3(x,y), I4(x,y) are the light intensity values ​​of the four-step phase-shifted fringe images respectively.

3. The three-coordinate structured light splicing system according to claim 1, characterized in that: The acquisition frame rate of the industrial camera is synchronized with the pattern projection rate of the structured light projection device, and the acquisition exposure time of a single image is adaptively adjusted by the ambient light intensity. The adjustment formula is: Where: t exp is the exposure time of a single image after adjustment; K is the exposure gain coefficient of the camera; L amb is the ambient light intensity detection value; b is the exposure bias coefficient of the camera; S is the image grayscale saturation threshold; The image grayscale saturation threshold S is preset by the system user.

4. The three-coordinate structured light splicing system according to claim 1, characterized in that: The preset trajectory of the motion control module adopts a curvature-based path planning algorithm to automatically adjust the scanning step size according to the curvature change of the workpiece surface. The curvature calculation formula is: Where: f″(x) is the second-order derivative of the function expression of the workpiece surface curve, which represents the rate of change of the curvature of the point on the curve; f′(x) is the first derivative of the function expression of the workpiece surface curve, which represents the slope of the tangent line at the point on the curve; When the curvature k is greater than a preset threshold, the scanning step is automatically reduced to half of the standard step.

5. The three-coordinate structured light splicing system according to claim 1, characterized in that: The pre-processing module uses a bilateral filtering algorithm for denoising, and the filtering formula is: Where g(i,j) is the grayscale value of the pixel (i,j) after filtering; f(k,l) is the grayscale value of the pixel (k,l) in the original point cloud data; Ω is the filtering window; c(i,j,k,l) ​​is the spatial proximity weight; s(i,j,k,l) ​​is the grayscale similarity weight.

6. The three-coordinate structured light splicing system according to claim 1, characterized in that: The calibration conversion module adopts a multi-station calibration method based on a standard sphere, and fits the center coordinates (X) of the standard sphere in the structured light coordinate system by the least square method. s ,Y s ,Z s ) and the sphere center coordinates in the world coordinate system (X w ,Y w ,Z w ), establish the homogeneous transformation matrix T; Where: The homogeneous transformation matrix T contains the rotation matrix R and the translation vector t, which is used to compensate the motion error of the coordinate measuring machine through iterative optimization.

7. The three-coordinate structured light splicing system according to claim 1, characterized in that: In the coarse registration stage of the stitching module, a feature matching algorithm based on FAST corner detection and FREAK feature descriptor is used, and false matching point pairs are eliminated through Lowe's ratio test. The matching efficiency is improved by more than 30% compared with the traditional SIFT algorithm. In the fine registration stage of the stitching module, an improved ICP algorithm is used to control the registration root mean square error within 0.005 mm by introducing a point cloud normal vector constraint term.

8. The three-coordinate structured light splicing system according to claim 1, characterized in that: The surface reconstruction algorithm of the output module adopts surface fitting based on the moving least squares method. By constructing a local quadratic polynomial to approximate the point cloud surface, the generated triangular mesh model satisfies the Euler formula: V-E+F=2(1-g); Where: V is the number of vertices; E is the number of edges; F is the number of faces; g is the mesh defect; Among them, the maximum ratio of the side length to the height of the grid is less than 8, and the minimum internal angle is greater than 20 degrees.

9. The three-coordinate structured light splicing system according to claim 1, characterized in that: The acquisition module is interactively connected to the motion control module via a wireless network, the motion control module is interactively connected to the preprocessing module and the calibration conversion module via a wireless network, the calibration conversion module is interactively connected to the splicing module via a wireless network, and the splicing module is interactively connected to the output module via a wireless network.

10. A method for splicing structured light in three coordinates, the method being an implementation method of the system for splicing structured light in three coordinates according to any one of claims 1 to 9, characterized in that: include: Step 1: Start the structured light projection device to project the composite coded pattern. The industrial camera synchronously collects the deformation stripe image on the workpiece surface. A single acquisition covers a single measurement area, and the original image data of each measurement area and the point cloud data in the structured light coordinate system are output; Step 2: Based on the size and shape of the workpiece, the probe is driven to move along a preset path to complete the scanning of each area and form a multi-view point cloud data set; Step 3: De-noise, smooth, and resample the original point cloud to output the processed point cloud data of each area; Step 4: Establish a mapping between the structured light coordinate system and the world coordinate system through the coordinate system calibration and calibration conversion module, so that the coordinate conversion matrix of each area point cloud is converted from the structured light coordinate system to the world coordinate system; Step 5: By extracting and matching the geometric features of adjacent regions, the random sampling consensus algorithm is applied to complete the coarse registration, and then the iterative closest point algorithm is used to optimize the parameters to complete the fine registration; Step 6: Unify the precisely registered point cloud into the world coordinate system, fuse the overlapping areas to generate a 3D point cloud model, and further convert the 3D point cloud model into a triangular mesh model.

Images