Automatic measurement system and method for double-view-field structured light turbine blade

Abstract

According to the automatic measurement system and method for the double-view-field structured light turbine blade, the double-view-field structured light measurement system is constructed, automatic viewpoint planning is carried out based on a turbine blade CAD model, discretization processing is carried out on the turbine blade CAD model to obtain point cloud data, a normal vector clustering algorithm is used for carrying out region segmentation on point cloud, and the point cloud data are subjected to region segmentation; estimating a scanning visual angle direction and a viewpoint position based on a clustering result, calculating displacement of each axis by combining an inverse kinematics model of the three-axis motion platform, and controlling the three-axis motion platform to move to a preset pose; double-view-field structured light hybrid measurement and coordinate system unification are executed, point cloud data obtained through multiple times of measurement are subjected to precise registration, point clouds are unified into an approximate global coordinate system through pose information of a three-axis motion platform, precise registration is conducted on the point clouds through an iterative nearest point algorithm, the splicing precision is improved, and accumulative errors are reduced.

Description

Technical Field

[0001] This invention relates to the field of high-precision structured light measurement technology, and in particular to an automated measurement system and method for dual-field structured light turbine blades. Background Technology

[0002] Three-dimensional vision structured light measurement is widely used in aerospace, automotive industry and medical fields due to its advantages such as high efficiency, full-field measurement and high degree of automation.

[0003] As the application scenarios of aero-engine turbine blades continue to develop towards high temperature, high pressure, and high speed, their design and manufacturing are also constantly putting forward new technical requirements for high precision, high quality, and automated processing. The guided processing and inspection of turbine blades have high requirements for measurement accuracy and integrity. General measurement methods are difficult to achieve efficient acquisition of high-precision, detailed local features of turbine blade full morphology measurement.

[0004] Traditional turbine blade measurement methods primarily rely on contact coordinate measuring machines (CMMs). While these offer high accuracy, they suffer from drawbacks such as slow measurement speed, poor adaptability to complex surfaces, and low automation, making it difficult to meet the combined requirements of efficiency, flexibility, and automation in modern blade production. In recent years, non-contact 3D measurement technologies, especially structured light-based 3D vision measurement methods, have gradually become an important development direction for blade measurement due to their advantages such as high speed, high resolution, and non-destructive testing.

[0005] Structured light-based 3D measurement systems typically reconstruct the surface topography of an object by projecting a specific fringe pattern onto its surface and using multiple cameras to calculate stereo parallax. Existing research and systems often employ binocular structured light measurement schemes, with a typical approach combining phase-shifting fringe analysis with 3D stereo matching technology for high-precision object reconstruction. However, conventional structured light systems still face numerous challenges when dealing with complex workpieces such as turbine blades, which have large surface curvatures, severe occlusion, and large size ranges. For example, while large field-of-view measurement systems can cover the entire blade, their spatial resolution is limited, making it difficult to capture fine local structures; conversely, small field-of-view systems offer higher resolution, but their measurement range is limited, making it difficult to cover the overall contour. Therefore, achieving a balance between large range and high precision is a key technical bottleneck in the current application of structured light measurement technology for blade measurement.

[0006] Furthermore, most existing structured light measurement systems rely on static layout and manual operation for measurement point planning, lacking linkage with CAD models, resulting in low efficiency in the measurement process and failing to achieve true automation.

[0007] In conclusion, it is necessary to study a dual-field structured light intelligent measurement system with high measurement accuracy, good integrity, and high degree of automation.

[0008] The information disclosed in the background section is only intended to enhance the understanding of the background of the present invention, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A dual-field structured light automated measurement method for turbine blades, including:

[0011] Step S1: Construct a dual-field-of-view structured light measurement system, wherein a pair of short-focal-length lenses and an industrial camera form a large-field-of-view structured light binocular measurement system to acquire the overall profile of the turbine blades; a pair of long-focal-length lenses and an industrial camera form a small-field-of-view structured light binocular measurement system to capture local detail information; four industrial cameras and a projector are arranged on the same horizontal line and mounted on a three-axis motion platform.

[0012] Step S2: Automatic viewpoint planning is performed based on the turbine blade CAD model. Specifically, the turbine blade CAD model is discretized to obtain point cloud data. The point cloud is segmented into regions using a normal vector clustering algorithm. The scanning viewpoint direction and viewpoint position are estimated based on the clustering results. The displacement of each axis is calculated by combining the inverse kinematics model of the three-axis motion platform, and the three-axis motion platform is controlled to move to the predetermined pose.

[0013] Step S3: Execute dual-field-of-view structured light hybrid measurement and coordinate system one. In this step, control four industrial cameras to synchronously acquire phase-shifted fringe patterns, perform three-dimensional reconstruction of the large field-of-view measurement area, and perform local three-dimensional reconstruction of viewpoints that need to supplement local details using a small field-of-view system. Based on the principle of phase consistency, search for the left and right view pixel coordinates of the same phase information of the horizontal and vertical stripes in the small field of view in the large field-of-view system and perform three-dimensional reconstruction. Use the matching points of the same name in the local reconstruction (also called matching points in this invention) to solve the coordinate transformation matrix between the two sets of points, thereby realizing the fusion of the large and small field-of-view measurement data.

[0014] Step S4 involves accurately registering the point cloud data obtained from multiple measurements. This involves using the pose information of a three-axis motion platform to unify each point cloud image into an approximate global coordinate system, and using an iterative nearest-point algorithm to perform precise registration of the point cloud, thereby improving stitching accuracy and reducing cumulative errors.

[0015] In the aforementioned automated measurement method for dual-field structured light turbine blades, step S2 includes:

[0016] Perform local plane fitting or principal component analysis on the point cloud to calculate the normal vector;

[0017] Clustering and segmentation are performed based on changes in the normal angle using a region growing algorithm;

[0018] Each cluster region is determined by principal component analysis to determine its main normal direction as the measurement viewpoint direction.

[0019] In the aforementioned dual-field structured light turbine blade automated measurement method, step S3 includes three-dimensional reconstruction,

[0020] Use a phase-shifting algorithm to obtain the package phase;

[0021] Phase expansion is performed using the multi-frequency heterodyne method;

[0022] Utilizing the principle of epipolar matching for stereo matching;

[0023] The least squares method is used for 3D point cloud reconstruction.

[0024] In the aforementioned dual-field structured light turbine blade automated measurement method, the phase-shifting algorithm is used to solve the wrapping phase. (x,y) represents the wrapping phase value:

[0025]

[0026] Package phase heterodyne calculation:

[0027]

[0028] ,

[0029] Here, `round()` is the floor function. The three-frequency heterodyne phase is unwrapped and subjected to two phase expansion calculations to finally obtain the monotonic phase.

[0030] Phase corresponding point search using the epipolar matching method:

[0031]

[0032] Where F is a 3×3 matrix, called the fundamental matrix, which represents the mapping from pixels in the left camera's image plane to epipolar lines in the right camera's image plane.

[0033] After obtaining the coordinates of the corresponding points in the phase-matched images, 3D reconstruction is performed on all point pairs between the left and right images using the camera calibration results and the least squares principle.

[0034] The projection matrices of the camera and the right camera are respectively and , and Given the coordinates of a point in both images, the 3D coordinates of the object's point W are:

[0035]

[0036]

[0037]

[0038] Reconstruct all matching phase point pairs to obtain a 3D point cloud.

[0039] In the aforementioned automated measurement method for dual-field structured light turbine blades, step S3 includes:

[0040] Extracting horizontal and vertical phase maps from a small field-of-view structured light system;

[0041] Search for pixel coordinates in a large field-of-view structured light system that have the same phase as the corresponding pixel in a small field of view;

[0042] Precise matching points are obtained through subpixel interpolation;

[0043] Solve the coordinate transformation matrix of the large and small field-of-view measurement system using the matching point set.

[0044] In the aforementioned automated measurement method for dual-field structured light turbine blades, step S4 involves precise registration of the point cloud data obtained from multiple measurements, including:

[0045] The point cloud is assigned an initial pose for each measurement based on the motion parameters of the three-axis platform;

[0046] Transform all point clouds to an approximate global coordinate system;

[0047] The iterative nearest point algorithm is applied to further register the point cloud, eliminating measurement errors and overlapping redundancy.

[0048] In the aforementioned dual-field-of-view structured light turbine blade automated measurement method, all point clouds are uniformly transformed to an approximate global coordinate system, and the pose estimation is completed by the measurement equipment.

[0049]

[0050] In the formula, P i P represents the point cloud set after a single measurement, while P represents the point cloud set after multiple measurements and coarse registration.

[0051] In the aforementioned automated measurement method for structured light turbine blades with dual field of view, the three-axis motion platform includes a BC dual rotating axis and a Z translation axis. The origin of the measurement reference system is the center point of the measurement focal plane field of view, and the origin of the platform reference system is the intersection of the dual rotating axes. A transformation relationship from the workpiece reference system to the platform reference system is established through a homogeneous transformation matrix to achieve alignment of the measuring equipment with the planned viewpoint.

[0052] A measurement system for performing the aforementioned automated measurement method for dual-field structured light turbine blades includes,

[0053] A dual-field-of-view structured light measurement system, comprising,

[0054] Four industrial cameras,

[0055] Two short-focus lenses and two long-focus lenses,

[0056] A high-resolution projector,

[0057] The three-axis motion platform consists of four cameras and a projector arranged on the same horizontal line to form a dual-field structured light measurement system. A pair of short-focal-length lenses and cameras form a large field-of-view system, while another pair of long-focal-length lenses and cameras form a small field-of-view system, which is used to achieve mixed measurement of the global profile and local details of turbine blades.

[0058] The viewpoint planning module is used to generate scanning paths and measurement postures based on CAD models.

[0059] The data fusion module is used to unify the coordinates and fuse point clouds of measurement data from both large and small fields of view.

[0060] The point cloud registration module is used to achieve mechanical stitching and ICP fine registration of point clouds under different measurement poses.

[0061] The measurement system described above measures aircraft engine turbine blades.

[0062] Compared with the prior art, the present invention has the following advantages: it adopts a strategy that combines a large field-of-view binocular structured light system with a small field-of-view binocular structured light system to achieve global accuracy and local high integrity measurement of the structured light system, thereby improving the overall accuracy of turbine blade measurement and the detail integrity of local features. Attached Figure Description

[0063] Various other advantages and benefits of the present invention will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0064] In the attached diagram:

[0065] Figure 1 This is a flowchart of the technology of the present invention;

[0066] Figure 2This is a system structure diagram of the present invention;

[0067] Figure 3 This is a schematic diagram of the dual-field structured light of the present invention;

[0068] Figure 4 This is a structural diagram of the three-axis motion platform of the present invention;

[0069] Figure 5 This is a schematic diagram of the matching of corresponding points in the dual system of the present invention;

[0070] Figure 6 This is a schematic diagram of multi-camera stripe acquisition according to the present invention;

[0071] Figure 7 This is a schematic diagram of the horizontal and vertical phases of the present invention;

[0072] Figure 8 This is a point cloud diagram illustrating the overall morphology of a single blade according to the present invention.

[0073] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0074] Specific embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0075] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions are preferred embodiments for carrying out the invention; however, these descriptions are for the purpose of understanding the general principles of the specification and are not intended to limit the scope of the invention. The scope of protection of this invention is determined by the appended claims.

[0076] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. The accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0077] like Figures 1 to 8 As shown, the automated measurement method for dual-field structured light turbine blades includes the following steps:

[0078] Step S1: Construct a dual-field-of-view structured light measurement system. A pair of short-focal-length lenses and an industrial camera form a large-field-of-view structured light binocular measurement system to obtain the overall profile of the turbine blades. A pair of long-focal-length lenses and an industrial camera form a small-field-of-view structured light binocular measurement system to capture local detail information. Four industrial cameras and a projector are arranged on the same horizontal line and mounted on a three-axis motion platform.

[0079] Step S2: Automatic viewpoint planning is performed based on the turbine blade CAD model. Specifically, the turbine blade CAD model is discretized to obtain point cloud data. The point cloud is segmented into regions using a normal vector clustering algorithm. The scanning viewpoint direction and viewpoint position are estimated based on the clustering results. The displacement of each axis is calculated by combining the inverse kinematics model of the three-axis motion platform, and the three-axis motion platform is controlled to move to the predetermined pose.

[0080] Step S3: Execute dual-field-of-view structured light hybrid measurement and coordinate system one. In this step, control four industrial cameras to synchronously acquire phase-shifted fringe patterns, perform 3D reconstruction of the large field-of-view measurement area, and perform local 3D reconstruction of viewpoints that need to supplement local details using the small field-of-view system. Based on the principle of phase consistency, search for the left and right view pixel coordinates of the same phase information of the horizontal and vertical stripes in the small field of view in the large field-of-view system and perform 3D reconstruction. Use the matching point set of the local reconstruction to solve the coordinate transformation matrix between the two sets of point sets, and realize the fusion of the large and small field-of-view measurement data.

[0081] Step S4 involves accurately registering the point cloud data obtained from multiple measurements. This involves using the pose information of a three-axis motion platform to unify each point cloud image into an approximate global coordinate system, and using an iterative nearest-point algorithm to perform precise registration of the point cloud, thereby improving stitching accuracy and reducing cumulative errors.

[0082] The above embodiment employs a strategy that combines a large field-of-view binocular structured light system with a small field-of-view binocular structured light system to achieve global precision and local high integrity measurement of the structured light system, thereby improving the overall accuracy of turbine blade measurement and the detailed integrity of local features.

[0083] It is understood that this invention essentially employs a three-axis motion platform scanning viewpoint estimation method based on CAD viewpoint planning, combined with a dual-field-of-view structured light measurement system, thereby improving the integrity and automation of the structured light system and achieving efficient automated measurement of turbine blades. The registration method within this invention effectively realizes a data registration method for structured light measurement systems with the same phase but different fields of view. This clearly helps to achieve accurate transformation of the coordinate systems of different field-of-view binocular measurement systems, as well as precise mapping of global and local measurement data of turbine blades.

[0084] In a preferred embodiment of the dual-field-of-view structured light turbine blade automated measurement method, in step S2, the normal vector clustering algorithm includes,

[0085] Perform local plane fitting or principal component analysis on the point cloud to calculate the normal vector;

[0086] Clustering and segmentation are performed based on changes in the normal angle using a region growing algorithm;

[0087] Each cluster region is determined by principal component analysis to determine its main normal direction as the measurement viewpoint direction.

[0088] In a preferred embodiment of the dual-field-of-view structured light turbine blade automated measurement method, step S3 includes three-dimensional reconstruction.

[0089] Use a phase-shifting algorithm to obtain the package phase;

[0090] Phase expansion is performed using the multi-frequency heterodyne method;

[0091] Utilizing the principle of epipolar matching for stereo matching;

[0092] The least squares method is used for 3D point cloud reconstruction.

[0093] In a preferred embodiment of the dual-field-of-view structured light turbine blade automated measurement method, the phase calculation uses a phase-shifting algorithm. (x,y) represents the wrapping phase value:

[0094]

[0095] Package phase heterodyne calculation:

[0096]

[0097] ,

[0098] Here, `round()` is the floor function. The three-frequency heterodyne phase is unwrapped and subjected to two phase expansion calculations to finally obtain the monotonic phase.

[0099] Phase corresponding point search using the epipolar matching method:

[0100]

[0101] Where F is a 3×3 matrix, called the fundamental matrix, which represents the mapping from pixels in the left camera's image plane to epipolar lines in the right camera's image plane.

[0102] After obtaining the coordinates of the corresponding points in the phase-matched images, 3D reconstruction is performed on all point pairs between the left and right images using the camera calibration results and the least squares principle.

[0103] The projection matrices of the camera and the right camera are respectively and , and Given the coordinates of a point in both images, the 3D coordinates of the object's point W are:

[0104]

[0105]

[0106]

[0107] Reconstruct all matching phase point pairs to obtain a 3D point cloud.

[0108] In a preferred embodiment of the dual-field-of-view structured light turbine blade automated measurement method, step S3 includes the following steps:

[0109] Extracting horizontal and vertical phase maps from a small field-of-view structured light system;

[0110] Search for pixel coordinates in a large field-of-view structured light system that have the same phase as the corresponding pixel in a small field of view;

[0111] Precise matching points are obtained through subpixel interpolation;

[0112] Solve the coordinate transformation matrix of the large and small field-of-view measurement system using the matching point set.

[0113] In a preferred embodiment of the dual-field-of-view structured light turbine blade automated measurement method, step S4, which involves precise registration of the point cloud data obtained from multiple measurements, includes...

[0114] The point cloud is assigned an initial pose for each measurement based on the motion parameters of the three-axis platform;

[0115] Transform all point clouds to an approximate global coordinate system;

[0116] The iterative nearest point algorithm is applied to further register the point cloud, eliminating measurement errors and overlapping redundancy.

[0117] In a preferred embodiment of the dual-field-of-view structured light turbine blade automated measurement method, all point clouds are uniformly transformed to an approximate global coordinate system, and the pose estimation is completed by the measurement equipment.

[0118]

[0119] In the formula, P i P represents the point cloud set after a single measurement, while P represents the point cloud set after multiple measurements and coarse registration.

[0120] In a preferred embodiment of the dual-field-of-view structured light turbine blade automated measurement method, the three-axis motion platform includes a BC dual-rotation axis and a Z translation axis. The origin of the measurement reference system is the center point of the measurement focal plane field of view, and the origin of the platform reference system is the intersection of the dual rotation axes. A transformation relationship from the workpiece reference system to the platform reference system is established through a homogeneous transformation matrix to achieve alignment of the measuring equipment with the planned viewpoint.

[0121] A measurement system for performing the aforementioned automated measurement method for dual-field structured light turbine blades includes,

[0122] A dual-field-of-view structured light measurement system, comprising,

[0123] Four industrial cameras,

[0124] Two short-focus lenses and two long-focus lenses,

[0125] A high-resolution projector,

[0126] The three-axis motion platform consists of four cameras and a projector arranged on the same horizontal line to form a dual-field structured light measurement system. A pair of short-focal-length lenses and cameras form a large field-of-view system, while another pair of long-focal-length lenses and cameras form a small field-of-view system, which is used to achieve mixed measurement of the global profile and local details of turbine blades.

[0127] The viewpoint planning module is used to generate scanning paths and measurement postures based on CAD models.

[0128] The data fusion module is used to unify the coordinates and fuse point clouds of measurement data from both large and small fields of view.

[0129] The point cloud registration module is used to achieve mechanical stitching and ICP fine registration of point clouds under different measurement poses.

[0130] In a preferred embodiment of the measurement system, the object of measurement is an aero-engine turbine blade.

[0131] In one embodiment, the method includes,

[0132] S1, the steps for constructing a measurement system based on dual-field structured light include:

[0133] S11, such as Figure 3 As shown, the dual-field structured light scanning system consists of four identical industrial cameras, two short-throw lenses, two long-throw lenses, and a high-resolution projector. The four cameras and the projector are arranged on a base on the same horizontal line, and the base is mounted at a certain angle on the vertical translation axis of the three-axis motion platform. Figure 4This is a structural diagram of the three-axis motion platform of the present invention. The three-axis platform can be considered as consisting of a BC dual-rotation axis and a Z translation axis. After estimating the scanning viewpoint through a CAD model, it is necessary to calculate the three-axis displacement through three-axis inverse kinematics modeling, moving each axis to make the optical axis of the dual-field-of-view system coincide with the planned viewing angle, and the center of the field of view located at the planned viewpoint. The system mainly consists of a ground reference frame. Platform Reference System Workpiece reference system and measurement reference system Composition, in which the platform reference frame origin For the two rotating axes of the measurement platform and At the theoretical intersection point, the origin of the measurement reference frame. To measure the center point of the focal plane field of view.

[0134] The S12 employs a pair of short-focal-length lenses to form a large-field-of-view structured light binocular measurement system, responsible for acquiring the global contour to ensure overall accuracy.

[0135] The S13 employs a pair of telephoto lenses to form a small field-of-view structured light measurement system, responsible for capturing local detail information. A high-resolution projector is used to encode the surface of the object under test, ensuring that high-contrast striped images can be obtained under different fields of view.

[0136] S2, the viewpoint planning method based on the design model includes the following steps:

[0137] S21 directly samples the turbine blade CAD model to obtain a uniformly distributed design model point cloud.

[0138] S22 applies a clustering and segmentation algorithm based on curvature and normal vectors to the model point cloud.

[0139] S221, The local plane fitting method is used to calculate the normal vector of the point cloud. A plane fitted by least squares with the set of points in its neighborhood. It can be represented as:

[0140]

[0141] The normal vector of the plane can be considered as Let be the normal vector at that point, and the normal vector can also be obtained through principal component analysis. (Plane) Passing through the centroid of the point set, and the normal vector satisfies The smallest eigenvalue can be obtained by constructing the covariance matrix of the point set and performing eigenvalue decomposition. The corresponding vector is the normal vector of that point.

[0142] S222: Based on the surface normal variation of the point cloud, the leading edge, trailing edge, leaf base, and underside regions of the leaf are obtained by clustering. Using a region growing algorithm, the growth starts from the seed point with the smallest normal variation. By continuously searching for neighboring points and calculating the angle between the normal of the neighboring point and the normal of the seed point, if the angle is less than a predetermined smoothing threshold, the neighboring point is added to the cluster. This process is repeated for all points in the point cloud until the clustering is completed.

[0143] S223. Then, PCA principal component analysis is used to determine the corresponding normal angle for each point cloud clustering result. This normal angle can be considered as the measurement viewing angle direction of the blade at that location. The field of view plane is the fitted plane, and the focal plane of the measuring equipment needs to be adjusted to coincide with it during measurement.

[0144] S23, based on the transformation relationship of the three-axis motion reference frame, establishes the transformation relationship from the workpiece reference frame to the platform reference frame through a homogeneous transformation matrix. Let the translational distance of the three platforms be z, and the rotation angles be b and c, respectively, when the workpiece moves from the current measurement point to the next measurement point. The position of the measurement origin in the ground reference frame is... , , The optical axis points to , , Platform reference system origin The position in the ground reference frame is , , The measurement points in the ground reference frame are denoted as follows after transformation in the workpiece reference frame: , , The measurement direction is , , .

[0145] The measurement reference frame and the ground reference frame only have Z-axis motion, while the platform reference frame has rotational motion along the B and C axes to the ground reference frame. Based on the transformation characteristics of homogeneous coordinates, two transformation matrices can be obtained as follows:

[0146]

[0147] Expanding, we get:

[0148]

[0149] To align the measuring equipment with the planned viewpoint, the center of the focal plane of the measuring system must coincide with the measurement point, and the optical axis must be aligned with the angle of view.

[0150]

[0151] The actual measurement viewpoint has been obtained through planning, while the axis is provided by the three-dimensional measurement results, and other variables need to be obtained through calibration. After inputting these variables, the motion parameters of each of the three axes can be obtained.

[0152] During automated measurement, the scanning viewpoint estimated from the CAD model is input into the inverse motion model to calculate the displacement of the three axes. The blade is then moved to a predetermined pose using two rotating axes and one translation axis for measurement. During measurement, the pose of the clamped blade is adjusted using rotating axes 1 and 2, and the measurement height of the dual-field scanner is adjusted using the translation axis. The relative pose of the blade and the dual-field structured light system is adjusted according to the measurement pose estimated from the CAD model to ensure that the scanning is at the optimal viewpoint. Multiple scans are performed according to the planned path trajectory, and the scan results are saved sequentially as a point sequence.

[0153] S3, the method steps for dual-field structured light hybrid measurement and coordinate system include:

[0154] S31. During measurement, four cameras are used to simultaneously acquire phase-shifted fringe patterns. Multi-frequency heterodyne phase calculation and phase unwrapping are performed on the fringe patterns corresponding to the field of view to be reconstructed. After stereo matching, three-dimensional reconstruction is performed.

[0155] The phase of the wrapper can be solved using a phase-shifting algorithm. (x, y) represents the wrapper phase value:

[0156]

[0157] Package phase heterodyne calculation:

[0158]

[0159]

[0160] The round() function is used to round down the data. Unwrapping the three-frequency heterodyne phase requires two phase expansion calculations to finally obtain the monotonic phase.

[0161] Phase corresponding point search using the epipolar matching method:

[0162]

[0163] Where F is a 3 × 3 matrix, called the fundamental matrix. It represents the mapping from pixels in the left camera's image plane to epipolar lines in the right camera's image plane.

[0164] After obtaining the coordinates of the corresponding points in the phase-matched images, 3D reconstruction is performed on all point pairs between the left and right images using the camera calibration results and the least squares principle.

[0165] The projection matrices of the camera and the right camera are respectively and , and Given the coordinates of a point in both images, the 3D coordinates of the object's point W are:

[0166]

[0167]

[0168]

[0169] Reconstruct all matching phase point pairs to obtain a 3D point cloud.

[0170] S32, local information supplementation using small field-of-view structured light.

[0171] For viewpoints requiring supplementary scanning with a small field of view, detailed reconstruction of local information is achieved through binocular structured light phase demodulation, matching, and 3D reconstruction.

[0172] like Figure 5 As shown: Under ideal conditions, corresponding points represent the same physical point in multiple views. In the case of no occlusion, corresponding points can be found one by one in multiple views. Therefore, the phase map can be obtained by solving the fringe map synchronously acquired by the dual-field structured light system, and then the matching of corresponding points in the dual-field can be achieved through phase matching.

[0173] like Figure 6 and Figure 7 As shown: Based on the principle of phase consistency, the coordinates of corresponding points are searched for horizontal and vertical stripes acquired synchronously by four cameras; after scanning and reconstructing the small field of view, the horizontal and vertical phase information in the small field of view is used again to search the large field of view system to find the left and right view pixel coordinates with the same phase information as the horizontal and vertical stripes in the small field of view, and then perform 3D reconstruction; thus realizing the reconstruction and matching of the corresponding matching point sets in the large and small fields of view.

[0174] S33 utilizes the corresponding points from local reconstruction to solve the coordinate transformation matrix of the two sets of point sets, thereby achieving the fusion of the two sets of measurement system data.

[0175] The same set of matching points represents the same physical point in the world in two binocular systems. Therefore, it is also the same corresponding 3D point in the dual-field structured light system. Thus, the 3D pose transformation relationship RT between the various binocular systems can be solved by the same set of matching points.

[0176]

[0177] In the formula, Representing the same point in the large field of view, This represents the corresponding point in the small field of view.

[0178] The coordinate system is transformed by the transformation matrix calculated from the matching point set of the same name for the small field of view structured light, so as to realize the fusion of dual field of view scanning data.

[0179]

[0180] In the formula, This represents the set of matching points with the same name in a large field of view. This represents the set of matching points with the same name in a small field of view.

[0181] S4, the steps for accurate registration using large and small field-of-view measurement data include:

[0182] S41 applies a fixed initial pose to each point cloud image, centered on the acquisition viewpoint, and then transforms all point clouds to an approximate global coordinate system. This process does not rely on feature matching or overlapping regions; it is accomplished solely through pose estimation by the measurement device.

[0183]

[0184] In the formula, The point cloud set representing a single measurement. This represents the set of point clouds after multiple measurements and coarse registration.

[0185] S42 utilizes ICP precision registration to further improve the stitching accuracy of point clouds and reduce cumulative errors.

[0186] In one embodiment, such as Figure 1 , Figure 2 As shown, an automated measurement method for turbine blades using dual-field-of-view structured light includes a dual-field-of-view structured light scanning system and a three-axis motion platform. First, a three-axis kinematic model is constructed, and the coordinate systems of the structured light measurement and the three-axis motion platform are unified. Then, a viewpoint is generated using a known CAD model of the blade, and the scanning posture, movement path, and whether to perform small-field-of-view compensation scanning for the complete scan are planned. Subsequently, the blade posture is adjusted using a dual-axis turntable, and the height of the dual-field-of-view scanning system is adjusted synchronously using a motion displacement stage, ensuring the dual-field-of-view scanning system is at the optimal scanning angle. Next, a fringe pattern is acquired and 3D reconstructed based on the planned scanning viewpoint, and small-field-of-view compensation scanning and data stitching are performed depending on whether local data supplementation is needed. Finally, the data from multiple scans are precisely registered using mechanical stitching and ICP, and redundancy is removed and the data is fused.

[0187] like Figure 2As shown, the dual-field-of-view structured light scanning system comprises four identical industrial cameras, two short-throw lenses, two long-throw lenses, and a high-resolution projector. The four cameras and projector are arranged on the same horizontal line. A pair of short-throw lenses forms a large-field-of-view structured light binocular measurement system responsible for acquiring the global contour; a pair of long-throw lenses forms a small-field-of-view structured light measurement system responsible for capturing local detail information. The high-resolution projector encodes the surface of the object under test, ensuring high-contrast stripe images are obtained under different fields of view. During measurement, for viewpoints that do not require small field-of-view compensation scanning, only vertical fringe patterns are projected, and stereo matching is performed using epipolar search. For viewpoints that require small field-of-view compensation scanning, both horizontal and vertical fringe patterns are projected, and four cameras simultaneously acquire phase-shifted fringe patterns. Stereo mapping is performed using epipolar search, and 3D reconstruction is performed based on the least squares principle. For phase-corresponding points in the small field-of-view structured light measurement system and the large field-of-view structured light measurement system, the left and right view pixel coordinates of the same phase information of the horizontal and vertical fringes in the small field of view are searched in the large field-of-view system according to the principle of consistency between horizontal and vertical phases, and 3D reconstruction is performed. Then, the coordinate transformation matrix of the two sets of points is solved using the corresponding points of the local reconstruction, realizing detailed reconstruction of local features and data stitching, and completing the fusion of data from the two dual-field-of-view measurement systems.

[0188] like Figure 5 As shown, the matching method between the large-field-of-view structured light measurement system and the small-field-of-view structured light measurement system in a dual-field-of-view structured light scanning system is as follows: For corresponding points in two phase maps under the same field of view, matching is performed using only one set of fringe maps via the epipolar search method; corresponding points under different fields of view are determined using the horizontal and vertical phase maps of each system's main view; for measurement viewpoints requiring supplementary small-field-of-view scanning, point clouds are reconstructed using both large-field-of-view and small-field-of-view structured light measurements, and corresponding points are reconstructed after phase matching under different fields of view using the horizontal and vertical phase maps; the pose transformation matrix of the measurement results from different field-of-view structured light measurement systems is solved using these corresponding points. Based on the pose transformation matrix, the measurement results from the small-field-of-view structured light measurement system are transformed into the coordinate system of the large-field-of-view structured light measurement system.

[0189] like Figure 7 As shown, the horizontal phase map and the vertical phase map are used. When the corresponding points under different fields of view are determined using the horizontal phase map and the vertical phase map of each system's main view, the horizontal coordinates of the corresponding points are first obtained through the horizontal phase map. Then, it is determined whether the vertical phase under the current coordinates is consistent with the corresponding points. When both the horizontal phase and the vertical phase meet the threshold requirements, sub-pixel interpolation is then performed to obtain accurate coordinates, thereby achieving accurate registration of the corresponding points under different fields of view.

[0190] like Figure 8As shown, a large field-of-view structured light measurement method was used to acquire the overall morphology of the blade to ensure overall accuracy; a small field-of-view structured light measurement method was used to acquire local detailed features of the blade to ensure the integrity of local detailed features; a telephoto lens was used to magnify the details of the local field of view to achieve refinement and precise reconstruction of local features; and phase information of the same name was used for matching to ensure the accuracy of point cloud coordinate system transformation. Red represents the large field-of-view point cloud, and green represents the small field-of-view point cloud. It can be clearly seen that the small field-of-view point cloud is denser and the details of the film vents are more accurate.

[0191] In one embodiment, the system comprises four identical industrial cameras, two short-focal-length lenses, two long-focal-length lenses, a high-resolution projector, and a three-axis motion platform. The four cameras and projector are arranged on the same horizontal line. A pair of short-focal-length lenses forms a large-field-of-view structured light binocular measurement system responsible for acquiring the global contour; a pair of long-focal-length lenses forms a small-field-of-view structured light measurement system responsible for capturing local detail information. During automated measurement, viewpoint planning is first performed: the normal is calculated based on the known CAD model, and clustering regions are segmented using the normal and neighborhood information to estimate the scanning viewpoint. Two rotation axes and one translation axis are moved to a predetermined pose for measurement. During measurement, the four cameras simultaneously acquire phase-shifted fringe patterns, and 3D reconstruction is performed based on the least squares principle. For areas requiring local information supplementation in the small field of view, local reconstruction is first performed, and the phase consistency principle is used to search for the left and right view pixel coordinates of the same phase information of the horizontal and vertical fringes in the small field of view within the large-field-of-view system, and 3D reconstruction is performed. Then, the coordinate transformation matrix of the two sets of points is solved using the corresponding points from the local reconstruction, achieving the fusion of the two sets of measurement system data. Finally, through mechanical splicing and ICP, the measurement data from different angles were accurately registered.

[0192] In one embodiment, the method includes:

[0193] Step S1:

[0194] The hardware for an automated measurement method of structured light turbine blades with dual field of view consists of four identical industrial cameras, two short-focal-length lenses, two long-focal-length lenses, a high-resolution projector, and a three-axis motion platform.

[0195] Step S2:

[0196] Automated measurement based on viewpoint generation from CAD model and triaxial inverse kinematics model.

[0197] Step S3:

[0198] Dual-field structured light hybrid measurement and coordinate system I.

[0199] Step S4:

[0200] Precise registration of measurement data for different blade widths is achieved through mechanical splicing and ICP.

[0201] In the aforementioned automated measurement method for dual-field structured light turbine blades, step S1 includes:

[0202] Step S11: Build a dual-field structured light measurement hardware platform, consisting of four identical industrial cameras, two short-focal-length lenses, two long-focal-length lenses, a high-resolution projector, and a three-axis motion platform.

[0203] Step S12: A pair of short-focal-length lenses are used to form a large field-of-view structured light binocular measurement system, which is responsible for acquiring the global contour.

[0204] Step S13: A small field-of-view structured light measurement system is formed by a pair of telephoto lenses to capture local detail information.

[0205] In the aforementioned automated measurement method for dual-field structured light turbine blades, step S2 includes:

[0206] Step S21: The point cloud of the blade CAD model is obtained by discretizing it, and then the normal vector of the point cloud is obtained by local plane fitting. calculate,

[0207] Step S22: Based on the continuity of the angle between normals, clustering is performed using a normal-based clustering algorithm, and viewpoint estimation is performed based on the clustering results.

[0208] Step S23: Input the estimated viewpoint and viewing direction results into the inverse kinematics model, combine the current platform's geometry and coordinate system transformation relationship, calculate the corresponding three-axis displacement, and move each axis to the corresponding position to measure the turbine blades.

[0209] In the aforementioned automated measurement method for dual-field structured light turbine blades, step S3 includes:

[0210] In step S31, four cameras simultaneously acquire phase-shifted fringe patterns, and perform multi-frequency heterodyne phase calculation and phase unwrapping on the fringe pattern corresponding to the field of view to be reconstructed, followed by stereo matching and three-dimensional reconstruction.

[0211] Step S32: For viewpoints requiring supplementary scanning with a small field of view, detailed reconstruction of local information is achieved based on binocular structured light phase demodulation, matching, and 3D reconstruction.

[0212] Step S33: Solve the coordinate transformation matrix of the two sets of points using the corresponding points from the local reconstruction, and realize the fusion of the two sets of measurement system data.

[0213] In the aforementioned automated measurement method for dual-field structured light turbine blades, step S4 includes:

[0214] Step S41: Transform all point clouds across the entire area into an approximate global coordinate system by using different measurement poses.

[0215] Step S42: Using ICP fine registration, the stitching accuracy of the point cloud is further improved and the cumulative error is reduced.

[0216] This invention's dual-field-of-view structured light measurement system consists of four industrial cameras, two short-focal-length lenses, two long-focal-length lenses, and a high-resolution projector. The short-focal-length lenses form a large-field-of-view binocular structured light system for acquiring the overall outline of the blade, while the long-focal-length lenses form a small-field-of-view binocular structured light system for capturing local details. All optical equipment is arranged on the same horizontal line and mounted on a three-axis motion platform to achieve multi-scale measurement. By combining large and small fields of view, it balances global accuracy with the integrity of local details, improving measurement efficiency and adaptability. One set of equipment can complete comprehensive measurements from macro to micro. It ensures consistent image quality; a unified light source and synchronous acquisition mechanism ensure consistent image contrast and clarity under different fields of view. It facilitates integration and automated operation; the modular design is conducive to system integration and automated control. Discrete sampling of the CAD model yields point clouds; normal vector clustering segmentation algorithm identifies regions such as the leading edge, trailing edge, leaf base, and leaf underside; principal component analysis is used to determine the optimal measurement viewing angle direction for each region; and the displacement of each axis is calculated using the inverse kinematics model of the three-axis platform to drive the platform to a predetermined pose. This system enhances measurement integrity by automatically generating optimal measurement paths based on geometric features, avoiding omissions of critical parts; reduces manual intervention by achieving full automation of the measurement process and lowering reliance on operator experience; improves measurement accuracy by ensuring each measurement is within the optimal viewing angle and focal length range; and supports measurement of complex curved surfaces, making it particularly suitable for workpieces with complex free-form surfaces such as turbine blades. It projects multi-frequency phase-shifting fringe patterns; extracts the enclosed phase using a phase-shifting algorithm; expands the phase using multi-frequency heterodyne; performs stereo matching using epipolar search; and reconstructs 3D point clouds based on the least squares principle. This improves reconstruction accuracy and stability; compared to traditional single-frequency phase-shifting methods, multi-frequency heterodyne effectively eliminates phase ambiguity and improves reconstruction reliability; it adapts to large fields of view and complex surfaces, especially suitable for processing workpieces with occlusion and high curvature areas such as blades; it supports high-resolution measurements, providing high-quality initial data for subsequent reconstruction of local details in small fields of view; and it enhances anti-interference capabilities, exhibiting strong robustness to non-ideal conditions such as ambient light and reflections. The small field-of-view and large field-of-view data fusion technology (based on corresponding point matching) extracts horizontal and vertical phase maps in the small field-of-view system; searches for pixel coordinates with the same phase information in the large field-of-view system; obtains precise matching points using sub-pixel interpolation; and solves the coordinate transformation matrix through the matching point set to achieve data fusion of small and large fields of view. This enables seamless data stitching, accurately aligning high-precision local details with the overall morphology; eliminates registration errors, avoiding the cumulative errors introduced by traditional calibration boards or manual registration; improves data consistency, ensuring that point clouds under different fields of view are expressed in a unified coordinate system; and supports refined detection, such as the accurate measurement and evaluation of microstructures like film holes and edge transition zones.The point cloud fine registration method based on mechanical stitching and ICP assigns an initial attitude to each measured point cloud according to the pose parameters of the three-axis platform; transforms all point clouds to an approximate global coordinate system; applies the ICP algorithm for further fine registration; eliminates measurement errors and overlapping redundancy, and improves the quality of the final point cloud. It enhances measurement continuity and integrity, supports efficient stitching of multiple scan results; reduces manual intervention, eliminating the need for feature extraction or manual point selection, achieving fully automatic registration; enhances data reliability, significantly improving the consistency and accuracy of the final point cloud through fine registration; and is suitable for complex geometric structures, especially for parts with complex shapes and varied surfaces such as blades. The three-axis motion platform and measurement reference system model adopts a three-axis structure with BC dual-rotation axis + Z translation axis; defines the ground reference system, platform reference system, workpiece reference system, and measurement reference system; establishes a homogeneous transformation matrix to describe the relationship between each reference system; and calculates the three-axis motion parameters through an inverse kinematics model to align the measurement system with the planned viewpoint. This system enables spatial linkage control of the measurement system, ensuring that the measuring equipment is always in the optimal measurement posture; it improves measurement flexibility and adaptability, supporting measurement needs at any angle and height; it simplifies the coordinate transformation process, facilitating the fusion and stitching of multi-view data through unified reference system modeling; and it enhances system scalability, providing a basic framework for future multi-sensor collaborative measurement.

[0217] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art can make many other forms based on the guidance of this specification and without departing from the scope of protection of the claims of the present invention, and all of these are within the scope of protection of the present invention.

Claims

1. An automated measurement method for dual-field structured light turbine blades, characterized in that, The method includes the following steps: Step S1: Construct a dual-field-of-view structured light measurement system, wherein a pair of short-focal-length lenses and an industrial camera form a large-field-of-view structured light binocular measurement system to acquire the overall profile of the turbine blades; a pair of long-focal-length lenses and an industrial camera form a small-field-of-view structured light binocular measurement system to capture local detail information; four industrial cameras and a projector are arranged on the same horizontal line and mounted on a three-axis motion platform. Step S2: Automatic viewpoint planning is performed based on the turbine blade CAD model. Specifically, the turbine blade CAD model is discretized to obtain point cloud data, the point cloud is segmented into regions, the scanning angle direction and viewpoint position are further estimated, and the displacement of each axis is calculated by combining the inverse kinematics model of the three-axis motion platform, and the three-axis motion platform is controlled to move to the predetermined pose. Step S3 involves executing dual-field-of-view structured light hybrid measurement and coordinate system one. This includes controlling four industrial cameras to simultaneously acquire phase-shifted fringe patterns, performing 3D reconstruction of the large field-of-view measurement area, and supplementing viewpoints requiring additional local details. A small field-of-view system is then used for local 3D reconstruction. Based on the principle of phase consistency, the left and right view pixel coordinates of the horizontal and vertical fringes with the same phase information in the small field of view are searched in the large field-of-view system and 3D reconstruction is performed. Finally, the coordinate transformation matrix between the two sets of points is solved using the matching point sets from the local reconstruction, achieving the fusion of the large and small field-of-view measurement data. Step S4 involves accurately registering the point cloud data obtained from multiple measurements. This involves using the pose information of a three-axis motion platform to unify each point cloud image into an approximate global coordinate system, and using an iterative nearest-point algorithm to perform precise registration of the point cloud, thereby improving stitching accuracy and reducing cumulative errors.

2. The automated measurement method for dual-field structured light turbine blades as described in claim 1, characterized in that, Preferably, step S2 includes, Perform local plane fitting or principal component analysis on the point cloud to calculate the normal vector; Clustering and segmentation are performed based on changes in the normal angle using a region growing algorithm; Each cluster region is determined by principal component analysis to determine its main normal direction as the measurement viewpoint direction.

3. The automated measurement method for dual-field structured light turbine blades as described in claim 1, characterized in that, In step S3, the three-dimensional reconstruction includes, Use a phase-shifting algorithm to obtain the package phase; Phase expansion is performed using the multi-frequency heterodyne method; Utilizing the principle of epipolar matching for stereo matching; The least squares method is used for 3D point cloud reconstruction.

4. The automated measurement method for dual-field structured light turbine blades as described in claim 3, characterized in that, The phase of the wrapping is solved using a phase-shifting algorithm. (x,y) represents the wrapping phase value. Stripe pattern intensity: ， Package phase heterodyne calculation, These are the phase values ​​of the two images, The result of the heterodyne calculation: ， ， Here, `round()` is the floor function. The three-frequency heterodyne phase is unwrapped and subjected to two phase expansion calculations to finally obtain the monotonic phase. , Phase corresponding point search using the epipolar matching method: ， Where F is a 3×3 matrix, called the fundamental matrix, which represents the pixels from the left camera imaging plane. To the right camera imaging plane The mapping of polar lines, After obtaining the coordinates of the corresponding points in the phase-matched images, 3D reconstruction is performed on all point pairs between the left and right images using the camera calibration results and the least squares principle. The projection matrices of the left and right cameras are respectively and , and Point coordinates in two images and Here, I represents the intrinsic parameter matrices of the left and right cameras, respectively, where I is the identity matrix, and R and T are the rotation and translation matrices. The 3D coordinates of the object's point W are: ， ， ， Reconstruct all matching phase point pairs to obtain a 3D point cloud.

5. The automated measurement method for dual-field structured light turbine blades as described in claim 1, characterized in that, Step S3 includes, Extracting horizontal and vertical phase maps from a small field-of-view structured light system; Search for pixel coordinates in a large field-of-view structured light system that have the same phase as the corresponding pixel in a small field of view; Precise matching points are obtained through subpixel interpolation; Solve the coordinate transformation matrix of the large and small field-of-view measurement system using a set of matching points with the same name.

6. The automated measurement method for dual-field structured light turbine blades as described in claim 1, characterized in that, In step S4, the precise registration of the point cloud data obtained from multiple measurements includes, The point cloud is assigned an initial pose for each measurement based on the motion parameters of the three-axis platform; Transform all point clouds to an approximate global coordinate system; The iterative nearest point algorithm is applied to further register the point cloud, eliminating measurement errors and overlapping redundancy.

7. The automated measurement method for dual-field structured light turbine blades as described in claim 1, characterized in that, The point cloud data is uniformly transformed to an approximate global coordinate system, and the pose estimation is accomplished through measurement equipment. ， In the formula, P i P represents the point cloud set after a single measurement, while P represents the point cloud set after multiple measurements and coarse registration.

8. The automated measurement method for dual-field structured light turbine blades as described in claim 1, characterized in that, The three-axis motion platform includes a BC dual-rotation axis and a Z translation axis. The origin of the measurement reference system is the center point of the measurement focal plane field of view, and the origin of the platform reference system is the intersection of the two rotation axes. The transformation relationship from the workpiece reference system to the platform reference system is established through a homogeneous transformation matrix to achieve alignment of the measuring equipment with the planned viewpoint.

9. A measurement system for performing the automated measurement method for dual-field structured light turbine blades as described in any one of claims 1-8, characterized in that, It includes, A dual-field-of-view structured light measurement system, comprising, Four industrial cameras, Two short-focus lenses and two long-focus lenses, A high-resolution projector The three-axis motion platform consists of four cameras and a projector arranged on the same horizontal line to form a dual-field structured light measurement system. A pair of short-focal-length lenses and cameras form a large field-of-view system, while another pair of long-focal-length lenses and cameras form a small field-of-view system, which is used to achieve mixed measurement of the global profile and local details of turbine blades. The viewpoint planning module is used to generate scanning paths and measurement postures based on CAD models. The data fusion module is used to unify the coordinates and fuse point clouds of measurement data from both large and small fields of view. The point cloud registration module is used to achieve mechanical stitching and ICP fine registration of point clouds under different measurement poses.

10. The measurement system as described in claim 9, characterized in that, The object of measurement is the turbine blade of an aircraft engine.

Images