A method and system for calibrating a single camera high precision scanning system

Abstract

The application provides a calibration method suitable for a single-camera high-precision scanning system, comprising the following steps: S1: linear array camera face array mode calibration; by adjusting the exposure time of the linear array camera, moving the base station to image at different positions of the checkerboard, solving the linear array camera internal parameter and distortion parameter; S2: linear scanning mode calibration; by imaging the linear scanning of the checkerboard, obtaining the pixel coordinates of the corner points of the checkerboard; separating the actual rotation angle of the checkerboard from the projection error, and then deriving the real camera rotation angle, so as to complete the calibration. The application proposes a two-stage algorithm combining face array mode modeling and linear scanning imaging correction for the calibration requirement of the linear array camera in the scanning imaging scene, and effectively solves the problem that the traditional calibration method is not applicable under the condition of single-camera fixation and single-direction motion.

Description

Technical Field

[0001] This application relates to the field of camera calibration technology, and in particular to a calibration method and system suitable for a single-camera high-precision scanning system. Background Technology

[0002] To achieve accurate mapping from image coordinates to spatial coordinates, any imaging-based measurement system requires geometric calibration of the camera to determine its intrinsic, extrinsic, and distortion parameters. Currently, common camera calibration methods solve for intrinsic and extrinsic parameters by forming regular patterns and quantitative transformation relationships. A representative method is the Zhang Zhengyou method, which observes a checkerboard or dot array from multiple perspectives and solves for the camera's intrinsic and extrinsic parameters and distortion coefficients based on corresponding points in world and image coordinates. A common characteristic of these methods is the need for multi-view observation; for example, the Zhang Zhengyou method requires three perspectives to solve for all intrinsic parameters, and each perspective has independent extrinsic parameters.

[0003] The aforementioned calibration algorithm has been widely applied in fields such as structured light measurement, stereo vision, and robot vision, and has achieved a high level of maturity. However, its applicability is significantly insufficient in high-precision scanning imaging scenarios (such as line scanning imaging in wafer defect detection), mainly in the following two aspects:

[0004] (i) Different imaging geometry: In line scan imaging, the camera position and orientation are fixed, and the platform moves along an approximately horizontal plane. This cannot meet the basic assumption of the diversity of external parameters in traditional methods, resulting in high uncertainty in the algorithm analysis process.

[0005] (ii) Significant differences in imaging physical processes: Line scan images are formed by pixel integration at different angles along the scanning direction. The imaging model differs significantly from that of conventional area scan cameras. Specific effects include:

[0006] 1. The imaging coordinate transformation is not a rigid transformation, and the orthogonality condition of the rotation matrix columns assumed by the conventional calibration algorithm no longer holds;

[0007] 2. Lens distortion in the scanning direction is smoothed out by integration, while distortion in the orthogonal direction is completely preserved. The distortion distribution is directional, which is different from the conventional model.

[0008] Conventional algorithms for line scan calibration are typically only applicable to low-resolution stereo scanning scenarios. A single viewpoint scan cannot calibrate the camera's intrinsic and extrinsic parameters. Therefore, these algorithms usually require additional conditions, such as multiple camera line scan angles, a 3D calibration object, assistance from a planar camera with known parameters, precise vertical movement of the platform or calibration object, and perfectly orthogonal placement of the checkerboard pattern. However, in high-precision industrial scanning scenarios, cameras and platforms are extremely delicate and fragile, making the above conditions unsuitable.

[0009] 1. Even slight changes in temperature and humidity can cause unacceptable system errors. After optical correction is completed, the optical system should be avoided as much as possible. Therefore, multi-degree scanning or vertical lens movement is not feasible.

[0010] 2. Imaging resolution is usually in the submicron level; therefore, in order to avoid defocusing and punctuation errors, the size of the three-dimensional calibration object needs to be in the micron level, and the process accuracy needs to be in the submicron level; even if cost is not a concern, it is difficult to achieve with existing processes.

[0011] 3. Typically, the substrate cannot and does not need to move vertically. Even substrates that can move vertically cannot achieve submicron level precision movement.

[0012] 4. To achieve submicron level imaging, extremely high-energy light sources and precision lenses are required, costing over a million; therefore, adding auxiliary lenses of the same precision is extremely expensive.

[0013] 5. Algorithms that do not require the above restrictions, such as the Donne algorithm, can obtain the camera intrinsic parameters, but cannot separate the rotation angle of the chessboard layout, i.e., the accurate extrinsic parameters; and its default is camera movement rather than the base, which is different from high-precision scanning scene modeling.

[0014] In summary, there is currently no suitable calibration algorithm for high-precision scanning systems. Therefore, traditional area scan camera calibration methods are still commonly used in high-resolution line scan scenarios. Because aberrations are significant in high-precision scanning and are often corrected through fitting, errors caused by area scan calibration algorithms may be reduced during aberration correction and are therefore often ignored. However, this conventional approach undoubtedly increases system errors and the difficulty of aberration correction. Summary of the Invention

[0015] In view of this, the purpose of this application is to propose a calibration method and system suitable for a single-camera high-precision scanning system, which can specifically solve the existing problems.

[0016] This application aims to solve the calibration problem of linear scan cameras in scanning scenarios, and is particularly suitable for applications where the camera position is fixed and the platform moves along a plane. Specifically, it includes:

[0017] 1. How to accurately determine the camera's magnification and optical axis deviation without multi-angle observation conditions;

[0018] 2. How to restore the relative rotation angle between the camera and the base, and eliminate image misalignment caused by the slight angle between the base's trajectory and the imaging direction;

[0019] 3. How to establish a distortion model suitable for line scanning mechanisms and accurately extract distortion parameters from scanned images.

[0020] Based on the above objectives, this application proposes a calibration method suitable for a single-camera high-precision scanning system, comprising:

[0021] S1: Linear Scan Camera Area Scan Mode Calibration

[0022] By adjusting the exposure time of the line scan camera and moving the platform to image at different positions on the checkerboard, the intrinsic parameters and distortion parameters of the line scan camera are solved.

[0023] S2: Line scan mode calibration

[0024] By scanning the chessboard grid lines, the pixel coordinates of the corner points are obtained; the actual rotation angle of the chessboard grid is separated from the projection error. This leads to the deduction of the actual camera rotation angle. This completes the calibration.

[0025] In summary, the advantages of this application and the user experience it brings to the calibration of single-camera high-precision scanning systems are as follows:

[0026] This application addresses the calibration requirements of line scan cameras in scanning imaging scenarios by proposing a two-stage algorithm that combines area array mode modeling and line scan imaging correction, effectively solving the problem that traditional calibration methods are not applicable under conditions of single camera fixation and unidirectional motion.

[0027] This application constructs a mathematically complete linear array camera calibration algorithm, with each step solvable linearly, applicable to scanning imaging scenarios involving a fixed camera and a shifting platform. The algorithm constructs a joint normalization matrix to model the geometric relationships of multiple images, and then directly calculates the liberation factor, rotation angle, and distortion parameters through linear transformation. This avoids the complex process of nonlinear optimization relied upon in traditional methods, exhibiting good accuracy and interpretability. Although this application allows the introduction of nonlinear optimization to improve accuracy, all core parameters can be solved iteratively analytically within a linear system, demonstrating both structural mathematical rationality and engineering feasibility. This scheme is particularly suitable for applications requiring extremely high calibration accuracy and model robustness, such as submicron-level wafer defect detection systems. Attached Figure Description

[0028] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments disclosed in this application and should not be construed as limiting the scope of this application.

[0029] Figure 1 A schematic diagram of a system using checkerboard calibration according to an embodiment of this application is shown (assuming that the Z-axis of each coordinate system is parallel).

[0030] Figure 2 A flowchart illustrating a calibration method for a single-camera high-precision scanning system according to an embodiment of this application is shown.

[0031] Figure 3 The diagram shows the range of the chessboard captured by the area array and line scan in the simulation experiment (the capture range is calculated based on the simulation settings, and in actual applications it can only be estimated by the parameters obtained through calibration).

[0032] Figure 4 The diagram shows a comparison between the wafer coordinates obtained by the algorithm in this application and the OpenCV-Zhang algorithm after performing a simulated linear scan of the wafer and the actual coordinates.

[0033] Figure 5 The diagram shows a comparison between the wafer coordinates obtained by the algorithm of this application and the Donne algorithm and the actual coordinates after performing a simulated linear scan of the wafer.

[0034] Figure 6 The diagram shows the XY axis reprojection error of the algorithm in this application after performing a simulated linear scan of the wafer (error is defined as the reprojected coordinates minus the true coordinates).

[0035] Figure 7 A configuration diagram of a calibration system suitable for a single-camera high-precision scanning system according to an embodiment of this application is shown.

[0036] Figure 8 A schematic diagram of the structure of an electronic device provided in one embodiment of this application is shown.

[0037] Figure 9 A schematic diagram of a storage medium provided in one embodiment of this application is shown. Detailed Implementation

[0038] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0039] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0040] The improvements in this application are as follows:

[0041] 1. Identify the inapplicability of traditional camera calibration methods in scanning imaging systems.

[0042] It is clearly pointed out that linear array calibration relies on multiple viewing angles and rigid models, and fails in online scanning systems due to changes in imaging structure and asymmetry in distortion direction; existing linear array calibration algorithms require dynamic adjustment of the camera or modification of the base, and laboratory algorithms are difficult to apply to industrial scenarios.

[0043] 2. A modeling method based on the "joint positive response matrix" for linear array camera area array mode is proposed.

[0044] The exposure time was adjusted to utilize a line scan camera for area scan shooting; all images were creatively unified into a set of parameters that shared rotation and height, retaining only translation as a viewpoint-related variable, thus constructing a new parameter decoupling method, reducing calibration errors while improving robustness.

[0045] 3. Construct a two-stage calibration process to decouple the checkerboard pattern from the camera rotation angle.

[0046] In the second stage, based on the single line scan image and the results of intrinsic parameter calculation, the rotation angle of the chessboard relative to the world coordinates is solved, and then the actual camera angle is deduced from the total rotation angle to compensate for the error caused by the inconsistency between the scanning direction and the movement direction.

[0047] 4. Possesses extensive scalability

[0048] This method is not only applicable to linear scanning systems, but can also be extended to other devices with fixed viewing angles and single-axis motion imaging structures (such as industrial inspection lines, scanning microscopes, etc.). The original structure of the system is maintained throughout the calibration process, which has versatility and engineering application value.

[0049] As mentioned above, current conventional algorithms are inadequate for calibrating high-precision industrial scanning. However, it's important to note that these algorithms are mostly developed in laboratory environments, with imaging systems often purchased as a whole, without addressing the fine-tuning of camera components such as lenses and photosensitive plates. In high-precision industrial scanning, due to the extremely high accuracy requirements, lenses and photosensitive elements often need to be custom-made and assembled separately. By simply adjusting the data readout method of the photosensitive system, we can enable line scan cameras to perform long-exposure imaging at the same location, a technique known as the area array imaging mode of line scan cameras. This mode combines line scan and area array calibration methods, overcoming the limitations of line scan calibration.

[0050] Therefore, this invention proposes a two-stage joint calibration algorithm for the calibration problem of high-precision line scanning systems. It only requires switching the exposure time of the photosensitive plate and does not require moving or changing the camera. This enables high-precision mapping from the original image to world coordinates in the line scanning imaging system, providing a reliable geometric basis for subsequent image stitching, reconstruction and defect detection.

[0051] The mathematical model for chessboard imaging is as follows:

[0052] Camera mathematical modeling in calibration describes the transformations between image coordinates and the world coordinate system. Extrinsic parameters describe the transformation from the world coordinate system to the camera coordinate system, while intrinsic parameters describe the transformation from the camera coordinate system to the image coordinate system. Here, we simply list the transformations between the measurable coordinate systems to describe the position of a point on the checkerboard in the image coordinate system after imaging. Furthermore, this application assumes that the ideal optical axis is parallel to the Z-axis of each extrinsic parameter coordinate system, thus ignoring level-related calibration errors.

[0053] like Figure 1 As shown, assume the chessboard squares are rotated by 90 degrees relative to the base. The camera rotates 2 degrees relative to the base. The checkerboard pattern and the base fit together perfectly, and the camera height is [missing information]. During the i-th area array shooting, the origin of the checkerboard coordinate system is taken as the angular point closest to the upper left angular point within the camera's field of view; for a angular point with index j in this shooting, the checkerboard coordinates are... Image coordinates are .

[0054] 1. External parameters of array mode

[0055] (1) From checkerboard coordinate system to base coordinate system

[0056] Using a homogeneous coordinate system, and since the Z-coordinate is 0, rows and columns containing only 0s can be removed; as mentioned above, the origin of the chessboard coordinate system is the corner closest to the top left corner within the field of view; assuming the origin of the chessboard coordinate system in the base coordinate system is... The lower corner of the chessboard coordinate system. The coordinates in the base coordinate system are The rotation angle of the chessboard is :

[0057] ;

[0058] (2) From checkerboard coordinate system to camera coordinate system

[0059] Assuming the base displacement during shooting is coordinates in camera coordinate system for

[0060] ,

[0061] Pick The above formula can be written as:

[0062] ;

[0063] 2. Area array mode distortion and camera intrinsic parameters

[0064] The distortion is calculated in the normalized plane; since the transformation process applies the same to all i,j indices, the indices are ignored; let the x-coordinate of the normalized plane before distortion be... After distortion, it becomes :

[0065] ,

[0066] in These are distortion parameters. Substituting them into the intrinsic parameter matrix yields the imaging coordinates, i.e., the coordinates in the image coordinate system. :

[0067] ,

[0068] in The focal lengths in the horizontal and vertical directions of the image. The horizontal and vertical coordinates of the main point.

[0069] 3. External parameters of line scan mode

[0070] In line scan mode, the coordinates of the base at the start of the line scan are taken as 0. The derivation process of the extrinsic matrix is ​​as follows:

[0071] ,

[0072] The final line scan mode transforms from the checkerboard coordinate system to the camera coordinate system as follows:

[0073] ;

[0074] 4. Line scan mode intrinsic parameters and distortion

[0075] In line scan mode, the camera scanning direction (i.e., the y-axis of the image coordinate system) is imaged through integration. Therefore, the magnification of the y-axis is entirely controlled by the stage movement speed. In actual scanning, to ensure image quality, it should be compared with... Consistent; and since the center of integration is always at the center of the optical axis, the distortion in the y-direction can be considered zero. Therefore, the distortion and intrinsic parameter model, i.e., the transformation from the camera coordinate system to the image coordinate system, can be expressed as:

[0076] .

[0077] Figure 2 A flowchart illustrating a calibration method for a single-camera high-precision scanning system according to an embodiment of this application is shown. Figure 2 As shown, the calibration method applicable to a single-camera high-precision scanning system includes:

[0078] Phase 1 S1: Linear Scan Camera Area Scan Mode Calibration

[0079] This stage aims to temporarily use the line scan camera as an area scan camera (i.e., keeping the camera stationary and increasing the exposure time for imaging); by adjusting the exposure time of the line scan camera and moving the platform to image at different positions on the checkerboard, the intrinsic parameters and distortion parameters of the line scan camera are solved. Specifically, it includes the following steps S11-S14:

[0080] S11. Solve for the joint normality matrix

[0081] The following linear system was constructed to solve for a joint positive response matrix that unifies the camera rotation angle and displacement height, while only varying the translational matrix; the minor effect of the tilt angle was ignored:

[0082] ,

[0083] in Let be the coordinates of the corner point in the chessboard coordinate system. For pixel coordinates, to Shared parameters for the rotated part (i.e.) ), Translate each image independently. Image sequence number The corner numbers within each image. Given the camera height, the observation equations for multiple images can be concatenated into a unified linear system on the right side of the matrix and the bottom of the vector, and the positive response matrix can be solved using least squares.

[0084] Note that since the base tilt angle can usually be accurately calibrated using a level, its influence is ignored in this algorithm to improve robustness. Therefore, the camera height is not considered. Without constraints, the camera height is obtained by measuring the distance to the camera base. (This height does not require high-precision measurement; its error can be compensated by the intrinsic parameters.) Substitute into the normalization solution to obtain... .

[0085] S12. Calculate the magnification and rotation angle using the normality matrix.

[0086] Based on the solution obtained in the previous step Based on the unit norm constraint and orthogonality constraint, construct the following linear system:

[0087] ,

[0088] Note that the above equation is for an overdetermined linear system, and the focal length magnification can be linearly solved using the least squares method. ;

[0089] Simultaneously solve for the rotation angle ,Notice It includes the coupling rotation angle between the checkerboard and the camera, so it needs to be further separated and cannot be directly used for image calibration.

[0090] .

[0091] S13. Solve for the distortion coefficient using magnification.

[0092] The above steps have yielded the result of the division. All intrinsic and extrinsic parameters. Let Take the value from the previous iteration (initially 0) and substitute it into the mathematical model described above:

[0093] For the distorted normalized coordinates, through the intrinsic parameter matrix and work out

[0094] For the normalized coordinates before distortion, through the extrinsic parameter matrix and work out

[0095] When the optical axis deviation is small, it can be solved using the following linear system. , :

[0096] ,

[0097] If the center deviation is large, it should be solved simultaneously. Construct the following linear system:

[0098] .

[0099] S14, Iterative Solution

[0100] Finally, iterate through the above three steps until the result stabilizes. In the first step of solving the normality matrix, the obtained intrinsic parameters and distortion parameters need to be used to inversely distort the image corner points before proceeding to complete the iterative convergence.

[0101] The distortion process can also be directly calculated using the Taylor approximation:

[0102] ,

[0103] The error is approximately Considering the physical definition of focal length magnification, which is equivalent to the number of rows and columns of the sensor at focal length, this error is negligible in most applications.

[0104] Compared with conventional area scan camera calibration algorithms, this algorithm unifies the camera rotation angle and displacement height, with only the translation being different; at the same time, it transforms different parts of the imaging process into linear problems through different constraints; thus solving the problem that line scan cameras only have a main direction viewpoint during the imaging process and usually lack enough angles to support full attitude recovery.

[0105] Phase 2 S2: Line scan mode calibration

[0106] This stage only requires one line scan to obtain the pixel coordinates of the chessboard corner points, which is sufficient to separate the actual rotation angle of the chessboard from the projection error. This leads to the deduction of the actual camera rotation angle. It does not depend on multi-angle images. The line scan speed is based on the magnification obtained from calibration. Adjustments are made to achieve line scan mode. Meanwhile, since tilt is not considered, in line scan mode .

[0107] First, using the previously defined internal parameters... Obtain the distortion-free normalized image coordinates in line scan mode. :

[0108] ,

[0109] Incorporate camera height Total rotation angle That is, camera rotation angle rotation angle with the chessboard sum ( )have

[0110] ,

[0111] Select two points on the chessboard that are far apart and of high quality (denoted by subscripts 1 and 2 in the following formula). We can obtain:

[0112] ;

[0113] Find the rotation angle of the chessboard square in the world coordinate system, thus completing the transformation from the composite angle. Decoupling from the actual line scan geometry.

[0114] Experimental verification:

[0115] As a verification, we established simulation experiments to compare the performance of the joint calibration algorithm proposed in this application (hereinafter referred to as the caliUni algorithm) with the classical area scan and line scan calibration algorithms:

[0116] 1. OpenCV's calibrateCamera algorithm. This function library is an improvement upon the well-known Zhang Zhengyou algorithm, with improvements including the introduction of the RANSAC strategy in solving positive definite matrices and support for intrinsic parameter initialization; it is the most stable and commonly used camera calibration library in the field of computer vision. (Hereinafter referred to as the OpenCV algorithm)

[0117] 2. The algorithm used in the 2017 paper "ROBUST PLANE-BASED CALIBRATION FORLINEAR CAMERAS" by Simon Donne et al. This algorithm is an improvement on the one published by Jamil Drareni et al. in 2010, and can directly deduce camera parameters from the line scan calibration object. As mentioned above, among existing line scan calibration algorithms, only this type of algorithm can be used, albeit with some difficulty, for high-precision scanning scenarios. (Hereinafter referred to as the Donne algorithm)

[0118] We will select a set of optical imaging systems commonly used in high-precision scanning and set parameters that closely match real-world installation and industrial error levels, such as distortion parameters, camera rotation angle, checkerboard rotation angle, and lens optical axis deviation. Our simulation process is as follows: constructing the checkerboard – area array calibration – line scan calibration – constructing target points on the scanning wafer – line scan simulation – reprojecting the line scan image.

[0119] The simulation environment is built based on a scenario of line scan defect detection on wafers, and incorporates some unavoidable system defects: lens magnification. The distortion parameters are [18.05, 17.95]. The optical axis offset is [10, 0.1]. [4,2] pixels; focus height The image sensor has a pixel size of 19.5 mm; the line scan imaging sensor has a pixel size of 5 μm and a pixel count of [256, 6000], which is approximately 0.277 μm in resolution; each square of the chessboard is 50 μm in size and [3, 2] mm in size. In the experiment, we performed five area scans at random locations, ensuring that at least one complete row of chessboard squares was captured in each scan; and one line scan at a random location. Considering potential limitations in the imaging sensor's memory, the line scan distance was set to 4800 pixels. Simultaneously, to simulate the door positioning error, we added a Gaussian random error with a standard deviation of 0.1 pixels. The calibration scenario for the simulation experiment is as follows: Figure 3 As shown.

[0120] After calibration using a checkerboard pattern, we need to simulate actual wafer imaging. The true error of the calibration algorithm is the difference between the image coordinates reprojected onto the world coordinate system using the calibration parameters and the real-world coordinates. To distinguish it from the checkerboard coordinate system, we refer to the world coordinate system as the wafer coordinate system. The true error is the wafer reprojection error, and the algorithm accuracy error in the calibration step is called the checkerboard fitting residual. We simulate scanning a 150mm diameter wafer (6-inch wafer, the smallest size in practical applications). We assume each die on this wafer is 8*5 mm in size, with a 2mm gap between dies. In the simulation, we mark a coordinate point every 1mm along the edge of the die, starting from the front, for error calculation. Wafer scanning simulation can be found in [reference needed]. Figure 4 , Figure 5 .

[0121] Algorithm Comparison Results

[0122] 1. Comparison with OpenCV algorithm

[0123] As shown in Table 1, under simulation settings, the algorithm presented in this paper is several orders of magnitude more accurate than the OpenCV algorithm, i.e., the area matrix calibration algorithm based on Zhang Youzheng's algorithm improved in the general OpenCV library. Figure 4 As shown, the error generated by the OpenCV algorithm appears small on the entire wafer scale, but considering that chips are typically manufactured using sub-micron to nanometer-scale processes, errors larger than micrometers are fatal for extremely precise applications such as chip defect detection and localization. The caliUni algorithm proposed in this application, however, can achieve sub-micron accuracy even under the influence of Gaussian noise of 1 pixel standard deviation. Considering that each pixel in the simulation experiment represents a resolution of 277 nanometers, and the added Gaussian noise, the algorithm performs very accurately.

[0124] Table 1. Comparison of the proposed algorithm (caliUni algorithm) and the OpenCV calibration algorithm based on Zhang Youzheng's algorithm in the simulation experiment.

[0125]

[0126] It should be noted that the extrinsic parameters of the OpenCV algorithm, such as camera rotation, are calculated with high precision, proving that our improvement to unify the extrinsic parameters of the OpenCV algorithm and the algorithm for line scan aspect separation are feasible and effective. However, the distortion parameters and optical axis offset have relatively large errors. This is because these parameters are not obtained under unified extrinsic parameters, but rather under the assumption of independent extrinsic parameters. This further proves the rationality of the algorithm proposed in this application and the necessity of high-precision scan calibration. At the same time, the caliUni algorithm cannot completely fit the parameters related to nonlinear distortion, namely optical axis offset and distortion parameters, which require further improvement.

[0127] 2. Comparison with Donne's algorithm

[0128] Because the Donne algorithm cannot handle lens distortion and checkerboard rotation, in this experiment... All values ​​are set to 0, and noise is also set to 0. For example... Figure 5 As shown, the wafer coordinates obtained by the Donne algorithm differ significantly from the actual coordinates, which is visually apparent. The experiment demonstrates that, although both are line scan algorithms, the mathematical models for stage movement and camera movement are completely different. Scan calibration algorithms with default camera movement, similar to the Donne algorithm, cannot be used for high-precision scanning and are only suitable for applications where the stage moves but the camera cannot be touched. Specific errors are shown in Table 2.

[0129] Table 2 Comparison of the algorithm presented in this paper (caliUni algorithm) and the Donne algorithm in simulation experiments.

[0130]

[0131] The error in line-scan algorithms stems from the difference between the shear-like transformations in reality and the rotational transformations assumed by the algorithm, and this difference increases with the scanning distance. In wafer scanning scenarios, the diameter of the target wafer is commonly 300mm, while this experiment used a wafer with a radius of 100mm for simulation; therefore, the error would be larger in practical applications. Furthermore, in Donne's paper, the scanning scene size is almost identical to a checkerboard grid; when the calibration checkerboard is large enough, even with different mathematical models, a relatively approximate coordinate reconstruction can be obtained. However, in high-precision wafer scanning scenarios, the long side length of the checkerboard is typically only on the order of one-hundredth of the scanning scene, further amplifying the algorithm's shortcomings.

[0132] Furthermore, it should be noted that the error of the caliUni algorithm in this experiment is larger than that in the OpenCV comparison experiment above (2.1µm vs 0.23µm). This phenomenon may be puzzling to readers because this experiment does not include distortion noise and is under more relaxed conditions. The reason for this phenomenon is that when the lens is distortion-free, the K-step calculation of the caliUni algorithm fails, resulting in a large deviation in the optical axis and distortion parameters; thus, the overall wafer coordinates are translated after reprojection. Most of the error of the caliUni algorithm comes from this translation error, which can be confirmed by the fact that its maximum error and average error are almost identical. This error does not affect applications such as mosaic defect recognition and can be eliminated by any known coordinate point on the wafer; more importantly, in practical applications, the lens will inevitably have distortion, so this problem does not exist.

[0133] Error Analysis

[0134] Under the experimental conditions described in the simulation experiment setup section (i.e., the OpenCV comparison experiment settings), we plotted the X and Y axis errors of the wafer reprojection error, where the error is the reprojected coordinates minus the true coordinates, as shown below. Figure 6 As shown, the error mainly occurs at both ends of the scan band (note that the scan direction is along the wafer's X-axis, such as...). Figure 3 (As shown). The X-axis error is less than 500 nanometers and is always positive, indicating that the algorithm still has translational errors in the scanning direction, which can be corrected later based on any determined coordinate point on the wafer. Ignoring translational errors, the X-axis error originates from the rotation, and the Y-axis error originates from the rotation, distortion coefficients, and optical axis deviation; however, the errors in the X and Y axes caused by the rotation originate from... and The former has a higher order and is more sensitive than the latter. The fact that the X-axis error is larger than the Y-axis error indicates that the overall error is mainly dominated by the angle calibration error; although lens-related calibrations such as distortion and optical axis deviate significantly from the true values, they are not the main sources of error.

[0135] The application provides a calibration system suitable for a single-camera high-precision scanning system. This system is used to execute the calibration method for a single-camera high-precision scanning system described in the above embodiments, such as... Figure 7 As shown, the system includes:

[0136] Line scan camera area array mode calibration module 401 is used to solve the intrinsic parameters and distortion parameters of the line scan camera by adjusting the exposure time of the line scan camera, moving the base to image at different positions of the checkerboard.

[0137] The line scan mode calibration module 402 is used to obtain the pixel coordinates of the corner points of the chessboard grid by performing a line scan image of the chessboard grid; and to separate the actual rotation angle of the chessboard grid from the projection error. This leads to the deduction of the actual camera rotation angle. This completes the calibration.

[0138] The calibration system for a single-camera high-precision scanning system provided in the above embodiments of this application and the calibration method for a single-camera high-precision scanning system provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.

[0139] This application also provides an electronic device corresponding to the calibration method for a single-camera high-precision scanning system provided in the foregoing embodiments, to execute the calibration method for a single-camera high-precision scanning system. This application does not limit the scope of the embodiments.

[0140] Please refer to Figure 8 This illustrates a schematic diagram of an electronic device provided by some embodiments of this application. For example... Figure 8 As shown, the electronic device 20 includes: a processor 200, a memory 201, a bus 202, and a communication interface 203. The processor 200, the communication interface 203, and the memory 201 are connected via the bus 202. The memory 201 stores a computer program that can run on the processor 200. When the processor 200 runs the computer program, it executes the calibration method for a single-camera high-precision scanning system provided in any of the foregoing embodiments of this application.

[0141] The memory 201 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 203 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.

[0142] Bus 202 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory 201 is used to store programs. After receiving an execution instruction, the processor 200 executes the program. The calibration method for a single-camera high-precision scanning system disclosed in any of the foregoing embodiments of this application can be applied to the processor 200, or implemented by the processor 200.

[0143] The processor 200 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 200 or by instructions in software form. The processor 200 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 201. The processor 200 reads the information in memory 201 and, in conjunction with its hardware, completes the steps of the above method.

[0144] The electronic device provided in this application embodiment and the calibration method for a single-camera high-precision scanning system provided in this application embodiment are based on the same inventive concept and have the same beneficial effects as the methods they employ, operate, or implement.

[0145] This application also provides a computer-readable storage medium corresponding to the calibration method for a single-camera high-precision scanning system provided in the foregoing embodiments. Please refer to... Figure 9 The computer-readable storage medium shown is an optical disc 30, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it executes the calibration method for a single-camera high-precision scanning system provided in any of the foregoing embodiments.

[0146] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0147] The computer-readable storage medium provided in the above embodiments of this application and the calibration method for a single-camera high-precision scanning system provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0148] It should be noted that:

[0149] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, this application is not directed to any particular programming language. It should be understood that the content of this application described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of this application.

[0150] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0151] Similarly, it should be understood that, in order to simplify this application and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of this application, various features of this application are sometimes grouped together into a single embodiment, figure, or description thereof. However, this method of disclosure should not be construed as reflecting an intention that the claimed application requires more features than are expressly recited in each claim. Rather, as reflected in the following claims, inventive aspects lie in fewer than all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of this application.

[0152] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0153] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of this application and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.

[0154] The various component embodiments of this application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the virtual machine creation system according to the embodiments of this application. This application can also be implemented as a device or system program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such an implementation of this application can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0155] It should be noted that the above embodiments are illustrative of this application and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several systems, several of these systems may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0156] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A calibration method suitable for a single-camera high-precision scanning system, characterized in that, Includes the following steps: S1: Linear Scan Camera Area Scan Mode Calibration By adjusting the exposure time of the line scan camera and moving the platform to image at different positions on the checkerboard, the intrinsic parameters and distortion parameters of the line scan camera are solved. S2: Line scan mode calibration By scanning the chessboard grid lines, the pixel coordinates of the corner points are obtained; the actual rotation angle of the chessboard grid is separated from the projection error. This leads to the deduction of the actual camera rotation angle. This completes the calibration. Step S1 specifically includes the following steps S11-S14: S11. Solve for the joint positive response matrix; S12. Calculate the magnification and rotation angle using the positive translation matrix; S13. Solve for the distortion coefficient using magnification; S14. Iterative solution; Step S11 includes the following steps: The following linear system solution is constructed, unifying the camera rotation angle and displacement height, and only translating the different joint positive response matrices; while neglecting the slight effect of the tilt angle: ， in Let be the coordinates of the corner point in the chessboard coordinate system. For pixel coordinates, to For parameters shared by the rotating parts, Translate each image independently; Image sequence number The corner numbers within each image. Given the camera height, the observation equations for multiple images can be concatenated into a unified linear system on the right side of the matrix and the bottom of the vector, and the positive response matrix can be solved using least squares. The camera height is obtained by measuring the distance to the camera base, and then substituted into the normalization solution to obtain... ; Step S12 includes: Through the solution Based on the unit norm constraint and orthogonality constraint, construct the following linear system: ， The above equation represents an overdetermined linear system. The focal length magnification can be linearly solved using the least squares method. ; Simultaneously solve for the rotation angle , Including the coupling rotation angle between the checkerboard pattern and the camera: 。 2. The method according to claim 1, characterized in that, In step S11, the obtained intrinsic parameters and distortion parameters are used to perform anti-distortion on the image corner points before solving the problem.

3. A calibration system suitable for a single-camera high-precision scanning system, using the method described in claim 1 or 2, characterized in that, include: The linear scan camera area array mode calibration module is used to solve for the intrinsic parameters and distortion parameters of the linear scan camera by adjusting the exposure time of the linear scan camera, moving the base to image at different positions on the checkerboard; The line scan mode calibration module is used to obtain the pixel coordinates of the corner points of the chessboard grid by performing a line scan image of the chessboard grid; and to separate the actual rotation angle of the chessboard grid from the projection error. This leads to the deduction of the actual camera rotation angle. This completes the calibration.

4. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor runs the computer program to implement the method as described in claim 1 or 2.

Images