Visual auxiliary interference detection optical path misalignment calibration method, alignment method and system

Abstract

The present application relates to the technical field of visual detection, and specifically provides a kind of visual auxiliary interference detection optical path misadjustment calibration method, alignment method and system, calibration plate and image acquisition device are arranged at four orthogonal axis positions of the workpiece to be measured respectively, by visual tool, only one constraint axis of camera coordinate system is parallel to the direction axis of the misadjustment to be measured of workpiece to be measured in the state, in this state, calibration plate is imaged under different illumination environment by image acquisition device, the projection point of scale line on calibration plate plane and fork wire are extracted;The actual physical coordinates of projection point and fork wire are calculated based on homography matrix, the local offset of corresponding axis position is obtained, the translation and rotation misadjustment of workpiece to be measured are obtained by linear decoupling of the local offset of four axis positions.Furthermore, according to the calibration result, alignment adjustment is carried out, the alignment system adopts closed-loop control, and a kind of calibration and alignment scheme with strong adaptability, high precision and convenient operation is provided for the interference detection of large aperture optical element.

Description

Technical Field

[0001] This invention belongs to the field of visual inspection technology, and in particular relates to a method, alignment method and system for calibrating optical path misalignment in visual-assisted interferometric detection. Background Technology

[0002] Interferometric testing is one of the high-precision measurement methods for evaluating the surface quality of workpieces. The effectiveness of this technology fundamentally depends on the strict alignment between the interferometric testing optical path and the workpiece under test. Traditional alignment methods rely entirely on the operator's experience, visually judging interference fringes and manually adjusting the mechanical mechanism, which has inherent drawbacks such as high subjectivity, low efficiency, difficulty in quantifying accuracy, and non-reproducibility.

[0003] To address the aforementioned issues, the industry has introduced machine vision technology to quantify alignment. In Chinese patent application CN118781185A, published on October 15, 2024, entitled "A Method and Optical System for Achieving Absolute Surface Detection Center Alignment Based on Sub-pixel Positioning of Visual Images," a method is proposed to drive the inspected element to rotate around an axis by multiple angles. The interferometer's imaging system acquires images of its physical edges, and through sub-pixel processing and geometric fitting, the position of the mechanical rotation center is deduced. Finally, alignment with the optical path is achieved by moving the workpiece. However, this approach has significant limitations, making it unsuitable for aligning large-diameter, stationary workpieces. First, it requires the workpiece to be precisely driven to rotate to generate data for calculation. Meter-level optical elements, due to their mass and assembly precision requirements, are considered immovable references during inspection, requiring adjustment of the optical path itself for alignment. Second, the features identified and measured are the physical edges of the workpiece body, intended to calibrate its motion trajectory; however, large-diameter workpieces require measurement of projected features characterizing the optical path's pose. Therefore, existing technologies cannot solve the specific engineering problem of "directly and accurately measuring and guiding the pose adjustment of the interference optical path for fixed large workpieces".

[0004] Furthermore, homography, as a tool in computer vision used to describe the projection transformation between two planes, is proposed in Chinese patent application CN104597722A, published on May 6, 2015, entitled "A Projection Overlay Alignment System and Alignment Method Based on Affine Transformation Homography Matrix." This invention proposes using the affine transformation homography matrix to quickly calculate the rotation and translation between the features of a mask image and a silicon wafer image, achieving rapid alignment in photolithography overlay. It solves the problem of relative pose matching between two known planar patterns (mask and silicon wafer). However, this method requires the object being measured to be a plane with a clear feature pattern, and its homography matrix is ​​used to directly calculate the transformation parameters between images, without establishing a mapping relationship between image features and an absolute physical coordinate system. Therefore, it cannot be directly used to measure the absolute positional deviation of the projected features of the optical path relative to a fixed workpiece reference. This method is mostly applied to specific scenarios such as photolithography and small-diameter workpieces, and cannot be applied to solve the high-precision alignment of meter-scale large-diameter optical components in non-orthogonal mounting postures in interferometric detection optical paths.

[0005] Currently, all solutions for interference detection alignment based on machine vision technology require strict constraints on the relative posture of the vision system and the workpiece under test. However, such posture constraints greatly increase the difficulty of on-site assembly and adjustment of the calibration alignment system. Moreover, since the assembly and adjustment cannot meet the posture constraints 100%, there will still be residuals. These residuals will introduce calibration errors, which in turn will lead to a decrease in alignment accuracy.

[0006] Therefore, there is an urgent need for a vision-assisted alignment method that can relax installation posture constraints, improve system adaptability, and maintain high alignment accuracy. Summary of the Invention

[0007] In view of this, the present invention aims to provide a visual-assisted interferometric detection method, alignment method and system for calibrating optical path misalignment. By designing a visual fixture that can relax camera mounting attitude constraints and introducing a homography matrix direct mapping model, the absolute coordinate measurement of the optical path projection features is realized. Finally, by combining multi-axis observation, the accurate translation and rotation misalignment is decoupled. The visual measurement overcomes the drawbacks of traditional visual alignment and effectively relaxes the constraints of camera mounting attitude, avoiding the problems of difficult installation and adjustment and residual errors that exist in traditional mechanical constraints.

[0008] To achieve the above objectives, the technical solution created by this invention is implemented as follows: The first aspect of this invention provides a method for calibrating optical path misalignment in visual-assisted interferometry detection, comprising: S1: Calibration plates are arranged at the four orthogonal axes of the workpiece to be tested, and four image acquisition devices are set accordingly. Based on the attitude description relationship of the left-handed coordinate system, the freedom constraint conditions of the image acquisition device at each axis are set as follows: the constraint axis of the image acquisition device and the direction axis of the misalignment to be measured of the workpiece to be tested are two skew lines on two parallel planes respectively; the constraint axis of the image acquisition device at each axis does not coincide with the direction axis to be measured at that axis in space, ensuring that the image acquisition device has rotational freedom about the corresponding constraint axis, and restricting other rotational freedoms that would change the direction of the constraint axis. S2: Use an image acquisition device to image the calibration plate, identify and extract the scribing lines and the calibration plate in the image, and calculate the projection points of the scribing lines on the plane of the calibration plate; Identify and extract the crosshairs of the interferometer on the calibration plate plane in the image; S3: Based on the pixel coordinates of the feature points on the calibration plate in the image coordinate system and the known physical coordinates of the feature points on the calibration plate in the physical coordinate system of the calibration plate, calculate the homography matrix between the calibration plate plane and the camera image plane; Based on the pixel coordinates of the projection point and crosshair in the image coordinate system of each axis of the workpiece under test, the physical coordinates of the projection point and crosshair in the physical coordinate system of the calibration plate are calculated using the homography matrix, and the vector difference between the projection point and crosshair in the physical coordinate system of the calibration plate is calculated as the local offset of the corresponding axis. S4: Based on the local offset of the four axes, linear decoupling is performed to obtain the translational and rotational misalignment of the workpiece under test, and the interference detection misalignment calibration is completed.

[0009] Preferably, the image acquisition device is a camera.

[0010] Preferably, the image acquisition device for each axis position is configured with the following degree-of-freedom constraints: The camera coordinate system is defined as follows: the Z-axis coincides with the optical axis of the camera lens, and the positive direction of the Z-axis points to the corresponding calibration plate along the optical axis. The YOZ plane is the pitch plane of the camera around the X-axis, and the XOZ plane is the yaw plane of the camera around the Y-axis, forming a left-handed coordinate system. The workpiece coordinate system is defined as follows: the Z-axis coincides with the optical axis of the workpiece to be measured, and the Y-axis is the direction axis of the misalignment of the workpiece to be measured, forming a left-handed coordinate system. The camera's torsional freedom around the Y-axis is restricted in the camera coordinate system, while its pitch freedom around the X-axis and rotation freedom around the Z-axis are allowed.

[0011] Preferably, the camera is fixed using a vision fixture, which includes: A workpiece back fixture set at the bottom of the workpiece to be tested to support the workpiece. And a visual sensing module for fixing the camera at each axis; Each visual sensing module includes: Rod A is horizontally fixed to the tooling surface on the back of the workpiece; Rod B is perpendicularly connected to rod A; Component C includes a camera mounting rod rotatably connected to rod B. The height of the camera mounting rod on rod B is adjustable, and it can rotate about the rotation axis connecting the camera mounting rod and rod B to achieve camera pitch angle adjustment. Rod B is also connected to a pitch adjustment and locking component for locking the camera pitch angle.

[0012] Preferably, in S2, under the first brightness environment, the calibration plate is imaged using an image acquisition device to obtain a first image, and the straight line equation of the scribe line and the plane equation of the calibration plate are identified and calculated in the first image. Under the second brightness environment, the calibration board is imaged using an image acquisition device to obtain a second image, and the pixel coordinates of the crosshairs are identified and calculated in the second image.

[0013] Preferably, the intersection point of the linear equation of the scribing line and the plane equation of the calibration plate is calculated as the projection point of the scribing line on the plane of the calibration plate, and the pixel coordinates of the projection point in the image plane are obtained.

[0014] Preferably, the transformation relationship between the pixel coordinates of the feature points on the calibration plate in the image coordinate system and the known physical coordinates of the feature points on the calibration plate in the physical coordinate system of the calibration plate is as follows: ; in, It is a non-zero scaling factor. For the camera intrinsic parameter matrix, This represents the pixel coordinates of the feature points on the calibration plate in the image coordinate system. This represents the known physical coordinates of the feature points on the calibration plate in the calibration plate's physical coordinate system. It is a homography matrix. Describes the first column vector of the rotation matrix. Let t represent the second column vector of the rotation matrix, and t represent the translation vector. Using the homography matrix, the pixel coordinates of the projected points in the coordinate system of each axis image of the workpiece under test are determined. and the pixel coordinates of the crosshairs Transformation to obtain the physical coordinates of the projection point in the real space plane The physical coordinates of the crosshairs in the real space plane The vector difference between the physical coordinates of the projection point and the physical coordinates of the crosshairs for each axis position is: = - ; in, This indicates the local offset of the corresponding axis position.

[0015] Preferably, the relationship between the local offset and misalignment of the four axes is established as follows: ; ; in, These represent the local offsets of the four axis positions, This represents the offset of the workpiece under test in the X-axis direction. This represents the offset of the workpiece under test in the Y-axis direction, and R represents the radius of the workpiece under test. This indicates the rotational offset of the workpiece under test around the optical axis; By decoupling the relationship between the local offsets and misalignments of the four axes, the translational misalignment of the workpiece under test can be obtained. , and rotation misalignment for: , .

[0016] The second aspect of this invention provides a visual-assisted interferometric detection optical path alignment method, comprising: obtaining the translational and rotational misalignment of the workpiece under test by a visual-assisted interferometric detection optical path misalignment calibration method; Based on the translational and rotational misalignment of the workpiece under test, the interferometric detection optical path or the workpiece under test is driven to compensate for the translational and rotational misalignment, thereby achieving alignment between the interferometric detection optical path and the workpiece under test.

[0017] The third aspect of this invention provides a visual-assisted interferometric detection optical path alignment system, comprising: The calibration module is configured to perform: obtaining the translational and rotational misalignment of the workpiece under test by using a visual-assisted interferometric detection optical path misalignment calibration method; The alignment module is configured to drive the interferometric detection optical path or the workpiece under test to compensate for the translational and rotational misalignments based on the translational and rotational misalignments of the workpiece under test. The calibration module and the alignment module form a closed-loop feedback control loop. After the alignment module performs compensation based on the translational and rotational misalignment, the calibration module re-acquires the image and updates the misalignment. The alignment module repeats the alignment operation until both the translational and rotational misalignment meet the preset alignment accuracy threshold.

[0018] Compared with the prior art, the present invention can achieve the following beneficial effects: This invention utilizes a high-precision calibration plate, a 2D industrial camera, and an image positioning algorithm to convert both the optical path projection features (crosshair spots) and the workpiece position reference (etched line projection points) into quantifiable values. This achieves the digitalization and automation of the alignment process, solving the problem that traditional interferometric alignment relies entirely on operators visually interpreting interference fringes and manually adjusting them. Furthermore, traditional visual alignment methods cannot quantitatively calibrate the misalignment, relying entirely on manual experience, resulting in unreproducible results and low accuracy. In contrast, the calibration and alignment method of this invention, verified through engineering practice, achieves repeatability accuracy better than 30μm and absolute measurement accuracy better than the laser displacement sensor standard by 10μm. The calibration results provide a traceable and reproducible quantitative benchmark for the alignment process.

[0019] Furthermore, compared to existing vision alignment schemes, which require strict parallelism between the camera image plane, the calibration plate plane, and the workpiece feature surface, placing extremely high demands on machining accuracy and on-site assembly and adjustment levels, making them difficult to implement in large workpiece inspection scenarios, this invention relaxes the attitude constraints. By introducing a homography matrix mapping model, it only requires constraining one coordinate axis of the camera coordinate system to be strictly parallel to the measured distance direction axis, allowing the camera to rotate freely around this axis. This reduces the installation requirement from "plane-to-plane parallelism" to "one-axis parallelism," significantly lowering the stringent requirements for mechanical assembly and adjustment. This allows the vision measurement device to adapt more flexibly and robustly to workpieces of different sizes and complex on-site environments, greatly reducing the difficulty of assembly and adjustment and the operational threshold, and significantly improving the engineering adaptability of the method.

[0020] This invention utilizes a homography matrix to directly map complex images containing distortion and perspective onto a physical plane, eliminating the need to generate or rely on intermediate corrected images. The algorithm outputs the absolute physical coordinates of feature points in one step, simplifying the processing chain and improving computational efficiency and reliability. This invention simultaneously meets the requirements for high-precision measurement while ensuring applicability and efficiency.

[0021] The present invention designs a matching vision fixture, including an A rod that can be adjusted along the back of the workpiece, a B rod that is perpendicular to the A rod, and a C component that can change the angle with the B rod, which together constitute an adaptive vision sensing module that can flexibly adjust the working height, horizontal distance and field of view of the camera.

[0022] Based on the calibration method, this invention further expands the closed-loop control in the alignment process, effectively overcoming the repetitive positioning error of the mechanical actuator and the residual deviation of single open-loop compensation, ensuring the high consistency and reliability of the final alignment accuracy. Attached Figure Description

[0023] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments and descriptions of the invention are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of a visual-assisted interferometric detection optical path misalignment calibration method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the vision tooling provided according to an embodiment of the present invention.

[0024] The reference numerals in the figures include: Image acquisition device 1, workpiece to be tested 2, component C 3, rod B 4, rod A 5, calibration plate 6, workpiece back fixture 7. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and do not constitute a limitation thereof. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the invention. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, some operations related to the invention are not shown or described in the specification. This is to avoid obscuring the core parts of the invention with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; the relevant operations can be fully understood based on the description in the specification and general technical knowledge in the art.

[0026] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined to form various implementations. Furthermore, the order of the steps or actions in the method description can be changed or adjusted in a manner readily apparent to those skilled in the art. Therefore, the various orders in the specification and drawings are merely for the clear description of a particular embodiment and do not imply a mandatory order, unless otherwise stated that a particular order must be followed.

[0027] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on this invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0028] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0029] The invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0030] Please see Figure 1 In one embodiment of the present invention, a method for calibrating optical path misalignment in visual-assisted interferometry detection is provided, comprising: S1: Calibration plates are arranged at the four orthogonal axes of the workpiece to be tested, and four image acquisition devices are set accordingly. Based on the attitude description relationship of the left-handed coordinate system, the degree of freedom constraint conditions of the image acquisition device at each axis are set as follows: the constraint axis of the image acquisition device and the direction axis of the misalignment to be measured of the workpiece to be tested are two skew lines on two parallel planes respectively; the constraint axis of the image acquisition device at each axis does not coincide with the direction axis to be measured at that axis in space, ensuring that the image acquisition device has rotational degree of freedom around the corresponding constraint axis, and restricting other rotational degrees of freedom that would change the direction of the constraint axis.

[0031] S2: Use an image acquisition device to image the calibration plate, identify and extract the scribing lines and the calibration plate in the image, and calculate the projection points of the scribing lines on the plane of the calibration plate; Identify and extract the crosshairs of the interferometer on the calibration plate plane in the image; S3: Based on the pixel coordinates of the feature points on the calibration plate in the image coordinate system and the known physical coordinates of the feature points on the calibration plate in the physical coordinate system of the calibration plate, calculate the homography matrix between the calibration plate plane and the camera image plane; Based on the pixel coordinates of the projection point and crosshair in the image coordinate system of each axis of the workpiece under test, the physical coordinates of the projection point and crosshair in the physical coordinate system of the calibration plate are calculated using the homography matrix, and the vector difference between the projection point and crosshair in the physical coordinate system of the calibration plate is calculated as the local offset of the corresponding axis. S4: Based on the local offset of the four axes, linear decoupling is performed to obtain the translational and rotational misalignment of the workpiece under test, and the interference detection misalignment calibration is completed.

[0032] Specifically, in step S1, the calibration device is initialized and set up, and attitude constraints are established. For example... Figure 2 As shown, calibration plates 6 are first set at the four orthogonal axes X+, X-, Y+, and Y- of the workpiece 2 to be tested, with each calibration plate 6 parallel to the workpiece 2. In this embodiment of the invention, mainly for large-aperture optical lenses (aperture greater than 1m) that are inconvenient to rotate, four lines are usually uniformly engraved on the side of the optical lens to mark its spatial position. These four lines correspond to the four orthogonal axes X+, X-, Y+, and Y- of the lens, respectively. That is, when installing the calibration plates 6, each calibration plate 6 is ensured to correspond to one line, and the line is perpendicular to the corresponding calibration plate 6. The lines on the side of the lens are achieved through machining. These are inherent pose marking lines of the optical lens itself, serving as a general position marker for optical processing.

[0033] After the calibration plate 6 is installed, an adaptive vision sensing system with relaxed attitude constraints needs to be built around the workpiece 2 using a vision fixture. Specifically, a matching vision fixture is designed to install the image acquisition device 1 to construct the adaptive vision sensing system. In this embodiment, the image acquisition device 1 uses an existing 2D industrial camera to image the calibration plate 6, the scribing lines, and the interferometer crosshairs. The vision fixture includes a vision sensing module and a workpiece back fixture 7. The workpiece back fixture 7 is a high-flatness platform set at the bottom of the workpiece 2 to support it. The vision sensing module is used to install the image acquisition device 1 and constrain its attitude. Four vision sensing modules are evenly arranged around the workpiece back fixture 7, each with a corresponding camera to image the calibration plate 6, scribing lines, and interferometer crosshairs at the corresponding axis. Each vision sensing module includes: component C 3, rod B 4, rod A 5, and image acquisition device 1. Rod A 5 is laterally connected to the surface of the workpiece back fixture 7, and its orientation is parallel to the workpiece back fixture 7. Each mounting position of rod A 5 corresponds to a calibration plate 6. A mounting interface is provided at the outer end of rod A 5; this interface is a pre-drilled positioning slot. Rod B 4 is inserted into the positioning slot and secured with fasteners. This mounting interface ensures a stable perpendicular orthogonal relationship between rod A 5 and rod B 4 in space, giving rod B 4 a definite and unchanging axial direction. Typically, rods A 5 and B 4 can be designed to rotate relative to each other, and the rotation angle of rod B 4 can be locked using fasteners. Component C 3, as the core carrier for camera mounting and orientation adjustment, typically adopts an integrated L-shaped bent plate structure to balance compactness, high rigidity, and multi-functional integration. Its structure can be divided into two clearly defined areas: a camera mounting base and an orientation adjustment arc surface. These correspond to the two parts of the L-shaped curved plate structure: the camera mounting rod and the arc rod, respectively. The camera mounting rod is a straight rod design, with a precision-machined mounting reference plane on its upper surface. This plane has positioning pin holes that match specific camera models, used to achieve a stress-free rigid connection with the camera body, ensuring that the camera's optical sensor plane maintains a defined pose relationship with the mounting base. The camera is mounted on this mounting reference plane through the positioning pin holes. The camera mounting rod is rotatably connected to rod B4 via rotating connecting parts such as a pivot. The height of the camera mounting rod on rod B4 is adjustable. Specifically, rod B4 can be designed to be telescopic, or multiple rotating connection points can be set on rod B4 to allow the camera mounting rod to be connected at different heights on rod B4.The arc rod is perpendicularly connected to the camera mounting rod at both ends. The side of the arc rod is a fan-shaped cylindrical arc surface. A 60° arc-shaped through groove is opened along the circumference of the fan-shaped cylindrical arc surface, which serves as a mechanical guide and range limiting mechanism for angle adjustment. A high-strength precision screw passes through the arc-shaped through groove and the U-shaped mounting groove on rod B4 in sequence. In addition, a lock nut matching the high-strength precision screw is also provided. After the high-strength precision screw passes through the arc-shaped through groove and the U-shaped mounting groove, it can be locked by the lock nut to fix the intersection position of the arc rod and rod B4. When camera pitch angle adjustment is required, the lock nut is loosened. There is no resistance between the arc-shaped groove and the screw, allowing the arc-shaped rod to slide along the arc of the rod, thus adjusting the camera's pitch. The curvature of the arc-shaped groove on the rod constrains the camera's pitch adjustment range. In this embodiment, the camera can be continuously adjusted within a range of 10° to 60°, adjusting the field of view for capturing feature information. After adjusting to the target angle, the lock nut is tightened. The axial tension of the screw generates significant static friction between the fan-shaped cylindrical surface of the arc-shaped rod and the mating surface of rod B4, locking component C3 at the desired angle, eliminating any gaps or movement, and meeting the stringent rigidity requirements of the measurement process. In this embodiment, the arc-shaped rod, screw, and lock nut together constitute a pitch adjustment and locking assembly connected to rod B4, used to release the camera's pitch adjustment capability and lock the camera's pitch angle. This design is not unique and can be replaced with other mechanical structures, as long as the camera's tilt angle can be adjusted.

[0034] The aforementioned visual fixture relaxes the traditional multi-plane parallel mechanical constraints, allowing the image acquisition device 1's degree of freedom constraints for each axis to be as follows: the constraint axis of the image acquisition device 1 and the misalignment direction axis of the workpiece 2 to be measured are two skew lines on two parallel planes, with no spatial intersection and no directional constraints, while allowing the camera to rotate around this axis within the fixture's adjustment range. This restricts the camera from generating torsional motion in the camera coordinate system.

[0035] like Figure 2 As shown in the figure, a more detailed explanation is provided using the coordinate systems in the diagram. First, the workpiece coordinate system and the camera coordinate system are defined, both of which are left-handed coordinate systems. For ease of description, the XYZ axes of the workpiece coordinate system and the camera coordinate system are distinguished and described separately. The XYZ axes of the workpiece coordinate system are represented by... , and This indicates that the XYZ axes of the camera coordinate system are represented by... , and This indicates that the workpiece coordinate system is defined. The axis coincides with the optical axis of the workpiece 2 to be measured. The plane is the optical plane of the workpiece 2 to be measured. The axis is the direction axis of the misalignment to be measured on the workpiece 2; the camera coordinate system... The axis coincides with the optical axis of the camera lens, and The positive direction of the axis points along the optical axis to the corresponding calibration plate 6. The plane is around the camera The pitch plane of the axis, The plane is around the camera The torsional plane of the axis. The degree of freedom constraints of the image acquisition device 1 at each axis position can be expressed as the requirement. and To maintain parallelism between skew planes with no intersection, in other words, the camera cannot rotate around them. Rotation restricts the left and right torsional degrees of freedom. This attitude constraint ensures that the pixel coordinates observed in subsequent images have a linear relationship with the actual physical space coordinates, avoiding non-linear relationships and increased computational complexity.

[0036] In step S2, the visual sensing device is constructed through the above steps. Cameras at each axis position acquire corresponding images to obtain image features of the scribing lines, calibration plate 6, and interferometer crosshairs. It should be noted that during the interferometric detection process, a crosshair exists at each axis position. These four crosshairs at the four axes are inherent features of the interferometric detection optical path, reflecting the spatial direction of the detection optical path.

[0037] Specifically, due to the varying clarity of the crosshairs, scribing lines, and calibration plate patterns under different lighting intensities, clear images of the workpiece's side scribing lines and calibration plate patterns can only be captured by a camera in a brighter environment. However, different optical elements have varying diffraction efficiencies in their corresponding diffraction elements (interferometric testing fixture CGH) during interferometric testing, resulting in different brightness levels for the crosshairs. Typically, the crosshairs are less bright, and capturing them in a brighter environment leads to unclear images, affecting the accuracy of subsequent calculations of the relative distance between the projection point and the crosshairs. Therefore, in this embodiment of the invention, a time-division acquisition strategy is employed. The camera captures images twice, under first and second brightness conditions, to obtain two types of clear images used to establish a reference and characterize the optical path state. Feature information is then fused within a unified coordinate framework.

[0038] First, in a high-brightness environment (where "high" refers to the ambient light intensity compared to the second shot, without specifying a particular light intensity, only ensuring that the scribe lines and calibration plate 6 can be clearly captured), the camera at each axis position is driven to capture images of the calibration plate 6 and the scribe lines within its corresponding field of view. The features of the calibration plate 6 (9 in this embodiment) can be identified in the first image. (11 checkerboard corner features), in this image, the calibration plate 6 is in a known "zero position" state that is parallel to the feature surface of the workpiece 2 to be measured, and the scribing lines in the figure will serve as the original reference benchmark for subsequent calculation of translation misalignment.

[0039] To achieve accurate conversion from image information to spatial geometric parameters, feature recognition and line finding of the etched lines are required. The goal is to correlate the etched line features on the side of the workpiece with a coordinate system based on the calibration plate, through geometric projection relationships. To accurately solve for the linear equations of the etched lines in the image plane, a standardized image preprocessing and feature extraction workflow is needed. Specifically, in the process of recognizing and extracting the etched lines and calibration plate from the image, two regions of interest are first selected in the first image; this process can be implemented using existing image recognition algorithms.

[0040] For the region of interest of calibration board 6 in the image, all corner points of calibration board 6 in the image are first extracted and identified to obtain the set of pixel coordinates of each corner point. These corner points of known size in the image provide image coordinate input data for subsequent homography matrix calculation.

[0041] For regions of interest (ROIs) within an image's indentation lines, edge detection is required to accurately extract the indentation line features. To overcome the influence of uneven illumination, background estimation and shadow correction can be performed on the ROI before edge detection. In the background estimation process, this embodiment employs Gaussian filtering or morphological opening to estimate the background illumination field of the ROI, thereby performing background correction. After background correction, the ROI needs to be binarized to separate foreground features from the background. In this embodiment, a local adaptive thresholding method is used to binarize the background-corrected ROI. After binarization, connected components in the image are extracted and analyzed. By performing binarized morphological processing on the connected components within the ROI containing the indentation line features, the regional information of the indentation line is obtained.

[0042] To further improve the accuracy of the positional information of the scribing lines in the image, this embodiment of the invention also performs sub-pixel-level edge localization on the two side boundaries of the coarse-precision scribing line region. Finally, the sub-pixel point set of the two side boundaries of the scribing line region is fitted with a straight line using the overall least squares method to obtain the straight line equation of the scribing line in the image plane. Further, based on the mechanical design constraint that "the side of the scribing line is strictly perpendicular to the calibration plate 6", this straight line equation is extended towards the calibration plate 6, and the intersection point with the plane equation of the calibration plate 6 is calculated. The intersection point is the "equivalent projection point of the workpiece engraving line on the calibration plate 6 plane", which serves as a fixed reference datum for the workpiece coordinate system in the measurement plane.

[0043] Obtain the pixel coordinates of the projection point Subsequently, with the camera and vision fixture postures strictly unchanged, in a low-light environment, images are captured again using the camera at each axis position to obtain a second image containing information on the misalignment characteristics between the detection optical path and the workpiece position. In this second image, the position of the crosshair emitted by the interferometer of the detection optical path is projected onto the feature point on the calibration plate; the crosshair position contains the current interferometric detection alignment misalignment information. After obtaining a clear second image, edge detection and sub-pixel-level edge positioning operations, similar to those used for the above-described scribe line processing, are used to accurately locate the pixel coordinates of the crosshair (usually represented by the crosshair centroid). Wherein, the pixel coordinates of the projection point The pixel coordinates of the crosshairs represent the projected position of the workpiece 2 under the current state. Characterizes the projection position of the interferometric detection optical path in the current state.

[0044] Since the spatial constraints did not change during the acquisition of the first and second images, the pixel coordinates of the projection points obtained in the first image are... , and the pixel coordinates obtained from the second image The calibration plate, scribing lines, and crosshairs are captured in two separate images, all within the same coordinate system. The main purpose is to clearly capture these elements to facilitate accurate feature extraction. In special circumstances, the calibration plate, scribing lines, and crosshairs can be clearly imaged and accurately identified and extracted simultaneously, or only one image can be captured.

[0045] In step S3, after feature acquisition and fusion in the image, a homography matrix is ​​constructed using the set of pixel coordinates of each corner point of the calibration board 6 in the image and the known physical coordinates of each corner point on the calibration board 6 in actual space. The homography matrix is ​​then used to calculate the misalignment based on the features in the image. Here, the homography matrix H is a... The matrix establishes the transformation relationship between the physical coordinates of any feature point on calibration plate 6 in the actual physical space and the pixel coordinates of the corresponding feature point on calibration plate 6 in the image coordinate system. The mathematical model of this transformation relationship is as follows: ; in, It is a non-zero scaling factor. For the camera intrinsic parameter matrix, This represents the pixel coordinates of the feature points on calibration plate 6 in the image coordinate system. This indicates the known physical coordinates of the feature points on calibration plate 6 in the calibration plate's physical coordinate system. It is a homography matrix. Describes the first column vector of the rotation matrix. Let t represent the second column vector of the rotation matrix, and t represent the translation vector.

[0046] By substituting the pixel coordinates of each corner point of calibration board 6 in the image and the known physical coordinates of each corner point of calibration board 6 in actual space into the mathematical model representing the transformation relationship, the optimal homography matrix can be solved by minimizing the reprojection error through the Direct Linear Transformation (DLT) algorithm combined with nonlinear optimization. .

[0047] After obtaining the homography matrix H, for any feature point in the image extracted in step S2, the pixel coordinates are... The physical coordinates of the homography matrix H on the calibration plate 6 plane in real physical space can be calculated. : .

[0048] in, It is the inverse of the homography matrix. It is a non-zero scaling factor. and These are physical coordinates that have not undergone scale normalization.

[0049] For the pixel coordinates of the projection points obtained in the image and the pixel coordinates of the crosshairs Substituting into the above formula, the physical coordinates of the projection point in the calibration plate's physical coordinate system can be calculated using the homography matrix H. The physical coordinates of the crosshairs in the calibration plate's physical coordinate system The vector difference between the projection point and the crosshair in the physical coordinate system of the calibration plate. = - , This refers to the local offset observed at that axis position.

[0050] In step S4, the local offset corresponding to each axis position is obtained. Subsequently, a system of equations was established based on the different response modes of translational and rotational misalignment for each axis position. The geometric principle behind the establishment of the system of equations is: two-dimensional translational misalignment between the workpiece under test and the detection optical path. This manifests as a "co-directional offset," meaning that for observation points at opposite ends of the same axis (such as X+ and X-), the offset caused by translational misalignment has the same sign; while the slight rotational misalignment around the optical axis (Z-axis)... This manifests as a "reverse offset," meaning that for observation points at both ends of the same axis, the offset caused by rotation has the opposite sign.

[0051] Therefore, the system of equations relating the local offsets and misalignments of the four axes is established as follows: ; ; in, These represent the local offsets of the four axis positions, This represents the offset of the workpiece under test in the X-axis direction. This represents the offset of the workpiece under test in the Y-axis direction, and R represents the radius of the workpiece under test. This indicates the rotational offset of the workpiece under test around the optical axis; By decoupling the relationship between the local offsets and misalignments of the four axes, the translational misalignment of the workpiece under test can be obtained. , and rotation misalignment for: , , ; Thus, the accurate and separate calibration of the displacement of multi-axis image features into the two-dimensional translational and rotational misalignment of the system was completed.

[0052] As an optional embodiment, a visual-assisted interferometric detection optical path alignment method is also provided. After obtaining the translational and rotational misalignment of the workpiece under test through the above-described visual-assisted interferometric detection optical path misalignment calibration method, further adjustments are made based on the translational misalignment of the workpiece under test. , and rotation misalignment The optical path or the workpiece under test is driven by mechanical equipment to detect the translational misalignment. , and rotation misalignment Compensation is performed to align the interferometric detection optical path with the workpiece under test.

[0053] As an optional embodiment, a vision-assisted interferometric detection optical path alignment system is also provided, including a calibration module and an alignment module, wherein the calibration module is configured to perform: obtaining the translational and rotational misalignment of the workpiece under test through the above-described vision-assisted interferometric detection optical path misalignment calibration method; the alignment module is configured to: drive the interferometric detection optical path or the workpiece under test to compensate for the translational and rotational misalignment based on the translational and rotational misalignment of the workpiece under test.

[0054] The calibration module and the alignment module form a closed-loop feedback control loop. After the alignment module performs compensation based on the translational and rotational misalignment, the calibration module re-acquires the image and updates the misalignment. The alignment module repeats the alignment operation until both the translational and rotational misalignment meet the preset alignment accuracy threshold.

[0055] In summary, the above description is merely a preferred embodiment of this specification and is not intended to limit the scope of protection of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this specification should be included within the scope of protection of this specification.

[0056] The systems, apparatuses, modules, or units described in one or more of the above embodiments may be implemented by a computer chip or entity, or by a product having a certain function. A typical implementation device is a computer. Specifically, a computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0057] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0058] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

Claims

1. A method for calibrating optical path misalignment in visual-assisted interferometric testing, wherein four scribed lines are provided on the side of the workpiece to be tested, characterized in that, include: S1: Calibration plates are arranged at the four orthogonal axes of the workpiece to be tested, and four image acquisition devices are set up accordingly. Based on the attitude description relationship of the left-handed coordinate system, the freedom constraint conditions of the image acquisition device at each axis are set as follows: the constraint axis of the image acquisition device and the direction axis of the misalignment of the workpiece to be measured are two skew lines on two parallel planes respectively; the constraint axis of the image acquisition device at each axis does not coincide with the direction axis to be measured at that axis in space, ensuring that the image acquisition device has rotational freedom about the corresponding constraint axis, and restricting other rotational freedom that would change the direction of the constraint axis; S2: Use an image acquisition device to image the calibration plate, identify and extract the scribing lines and the calibration plate in the image, and calculate the projection points of the scribing lines on the plane of the calibration plate; Identify and extract the crosshairs of the interferometer on the calibration plate plane in the image; S3: Based on the pixel coordinates of the feature points on the calibration plate in the image coordinate system and the known physical coordinates of the feature points on the calibration plate in the physical coordinate system of the calibration plate, calculate the homography matrix between the calibration plate plane and the camera image plane; Based on the pixel coordinates of the projection point and crosshair in the image coordinate system of each axis of the workpiece under test, the physical coordinates of the projection point and crosshair in the physical coordinate system of the calibration plate are calculated using the homography matrix, and the vector difference between the projection point and crosshair in the physical coordinate system of the calibration plate is calculated as the local offset of the corresponding axis. S4: Based on the local offset of the four axes, linear decoupling is performed to obtain the translational and rotational misalignment of the workpiece under test, and the interference detection misalignment calibration is completed.

2. The method for calibrating optical path misalignment in visual-assisted interferometry detection according to claim 1, characterized in that, The image acquisition device is a camera.

3. The method for calibrating optical path misalignment in visual-assisted interferometry detection according to claim 2, characterized in that, The camera coordinate system is defined as follows: the Z-axis coincides with the optical axis of the camera lens, and the positive direction of the Z-axis points to the corresponding calibration plate along the optical axis. The YOZ plane is the pitch plane of the camera around the X-axis, and the XOZ plane is the yaw plane of the camera around the Y-axis, forming a left-handed coordinate system. The workpiece coordinate system is defined as follows: the Z-axis coincides with the optical axis of the workpiece to be measured, and the Y-axis is the direction axis of the misalignment of the workpiece to be measured, forming a left-handed coordinate system. The camera's torsional freedom around the Y-axis is restricted in the camera coordinate system, while its pitch freedom around the X-axis and rotation freedom around the Z-axis are allowed.

4. The method for calibrating optical path misalignment in visual-assisted interferometry detection according to claim 3, characterized in that, The camera is fixed by a vision fixture, which includes: A workpiece back fixture set at the bottom of the workpiece to be tested to support the workpiece. And a visual sensing module for fixing the camera at each axis; Each visual sensing module includes: Rod A is horizontally fixed to the tooling surface on the back of the workpiece; Rod B is perpendicularly connected to rod A; Component C includes a camera mounting rod rotatably connected to rod B. The height of the camera mounting rod on rod B is adjustable, and it can rotate about the rotation axis connecting the camera mounting rod and rod B to achieve camera pitch angle adjustment. Rod B is also connected to a pitch adjustment and locking component for locking the camera pitch angle.

5. The method for calibrating optical path misalignment in visual-assisted interferometry detection according to claim 2, characterized in that, In step S2, under the first brightness environment, the calibration plate is imaged using an image acquisition device to obtain a first image, and the straight line equation of the scribe line and the plane equation of the calibration plate are identified and calculated in the first image. Under the second brightness environment, the calibration board is imaged using an image acquisition device to obtain a second image, and the pixel coordinates of the crosshairs are identified and calculated in the second image.

6. The method for calibrating optical path misalignment in visual-assisted interferometry detection according to claim 5, characterized in that, The intersection point of the linear equation of the scribing line and the plane equation of the calibration plate is calculated and used as the projection point of the scribing line on the plane of the calibration plate, thus obtaining the pixel coordinates of the projection point in the image plane.

7. The method for calibrating optical path misalignment in visual-assisted interferometric detection according to claim 1, characterized in that, The transformation relationship between the pixel coordinates of feature points on the calibration plate in the image coordinate system and the known physical coordinates of feature points on the calibration plate in the physical coordinate system of the calibration plate is as follows: ； in, It is a non-zero scaling factor. For the camera intrinsic parameter matrix, This represents the pixel coordinates of the feature points on the calibration plate in the image coordinate system. This represents the known physical coordinates of the feature points on the calibration plate in the calibration plate's physical coordinate system. It is a homography matrix. Describes the first column vector of the rotation matrix. Let t represent the second column vector of the rotation matrix, and t represent the translation vector. Using the homography matrix, the pixel coordinates of the projected points in the coordinate system of each axis image of the workpiece under test are determined. and the pixel coordinates of the crosshairs Transformation to obtain the physical coordinates of the projection point in the real space plane The physical coordinates of the crosshairs in the real space plane The vector difference between the physical coordinates of the projection point and the physical coordinates of the crosshairs for each axis position is: = - ； in, This indicates the local offset of the corresponding axis position.

8. The method for calibrating optical path misalignment in visual-assisted interferometry detection according to claim 7, characterized in that, The relationship between the local offset and misalignment of the four axes is established as follows: ； ； in, These represent the local offsets of the four axis positions, This represents the offset of the workpiece under test in the X-axis direction. This represents the offset of the workpiece under test in the Y-axis direction, and R represents the radius of the workpiece under test. This indicates the rotational offset of the workpiece under test around the optical axis; By decoupling the relationship between the local offsets and misalignments of the four axes, the translational misalignment of the workpiece under test can be obtained. , and rotation misalignment for: ， 。 9. A visually assisted interferometric detection optical path alignment method, characterized in that, The translational and rotational misalignment of the workpiece under test are obtained by the visual-assisted interferometric detection optical path misalignment calibration method as described in any one of claims 1-8. Based on the translational and rotational misalignment of the workpiece under test, the interferometric detection optical path or the workpiece under test is driven to compensate for the translational and rotational misalignment, thereby achieving alignment between the interferometric detection optical path and the workpiece under test.

10. A visually assisted interferometric detection optical path alignment system, characterized in that, include: The calibration module is configured to perform: obtaining the translational and rotational misalignment of the workpiece under test by means of the visual-assisted interferometric detection optical path misalignment calibration method as described in any one of claims 1-8; The alignment module is configured to drive the interferometric detection optical path or the workpiece under test to compensate for the translational and rotational misalignments based on the translational and rotational misalignments of the workpiece under test. The calibration module and the alignment module form a closed-loop feedback control loop. After the alignment module performs compensation based on the translational and rotational misalignment, the calibration module re-acquires the image and updates the misalignment. The alignment module repeats the alignment operation until both the translational and rotational misalignment meet the preset alignment accuracy threshold.

Images