A method and system for calculating parallelism of a detector dual-plane joint optimization

Abstract

The application provides a calculation method and system for parallelism of dual-plane joint optimization of a detector, and relates to the technical field of optical detection. The calculation method comprises: collecting the coordinates of the lower surface positioning points, the upper surface reference points and the pixel point cloud of the detector, and performing deduplication and coordinate calibration; constructing initial planes based on the upper and lower surface reference points respectively and fusing them into a dual-surface initial average trend surface; generating prior weights according to the distance of the pixels to the trend surface; taking the initial plane parameters as the initial values of iteration, iteratively solving the dual-plane collaborative optimization objective function under the guidance of the prior weights, and synchronously obtaining the upper and lower surface fine plane parameters; and finally calculating the angle between the normal vectors of the two planes to obtain the parallelism. The application effectively suppresses the interference of outliers through the dual-surface trend surface construction and collaborative optimization mechanism, avoids the error accumulation of step-by-step fitting, significantly improves the accuracy and robustness of the parallelism calculation, and is especially suitable for the precise assembly and adjustment of high-resolution optical detectors.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of optical detection, in particular to a method for calculating parallelism of dual-plane joint optimization of a detector and a system thereof. BACKGROUND

[0002] With the increasing demand for large field of view, high sensitivity and high resolution detection, single-chip detectors are difficult to achieve ultra-large array or ultra-high resolution due to limitations of material growth, manufacturing process and physical properties. To break through the limitations, splicing technology emerges as the times require. The surface flatness of a spliced detector is a key indicator that determines its performance, directly affecting detection accuracy, imaging quality and system reliability. The core of parallelism calculation is to fit the three-dimensional planes of the upper and lower surfaces of the detector and solve the included angle between the normal vectors of the two planes. Currently, the plane fitting technology in the prior art mainly faces the interference of point cloud data noise and outliers, and it is difficult to balance the anti-interference ability and fitting accuracy of the algorithm.

[0003] In related technologies, the technical solutions mainly include traditional weighted least squares fitting and robust fitting methods. The traditional weighted least squares fitting solves the plane by minimizing the weighted residual sum of squares by presetting fixed weights for different points in the point cloud, but the weights are fixed and cannot dynamically respond to outliers caused by sudden noise during measurement. If such outliers are given high weights, the fitted plane will deviate significantly. Huber robust fitting uses a segmented loss function to dynamically adjust the weights, and when the residual is greater than a certain threshold, the weight of the point is reduced, thereby suppressing the influence of outliers. However, this method only relies on the posterior information of the iterative residual and does not combine the prior knowledge of the physical position of the detector pixels, resulting in insufficient weight differentiation, and the fitting accuracy is still lower than expected in complex noise scenarios. Moreover, in the existing technical solutions, the expansion of the splicing scale leads to a multiplication of the number of positioning points and reference points, and the step-by-step fitting causes the error transmission chain across the surface to be lengthened, resulting in a non-linear amplification of the final parallelism error. Large array introduces global stress warping, and the upper and lower surfaces no longer satisfy the assumption that the single-surface topography can represent the whole, resulting in the failure of the single-surface outlier suppression strategy.

[0004] Therefore, a method for calculating the parallelism of dual-plane joint optimization of a detector and a system thereof are provided to solve one of the above technical problems. SUMMARY

[0005] The purpose of the present application is to provide a method for calculating the parallelism of dual-plane joint optimization of a detector and a system thereof, which can solve at least one of the above technical problems. The specific scheme is as follows:

[0006] According to the specific embodiment of the present application, a method for calculating the parallelism of dual-plane joint optimization of a detector includes the following steps:

[0007] acquiring coordinates of at least four positioning points on a lower surface of the detector, coordinates of at least three reference points on an upper surface of the detector, and coordinates of all pixel point clouds on the upper surface of the detector;

[0008] performing duplicate data removal and coordinate calibration on the coordinates of the reference points on the upper surface and the coordinates of all pixel point clouds on the upper surface;

[0009] calculating initial plane parameters of the lower surface and initial plane parameters of the upper surface respectively according to the positioning points on the lower surface and the reference points on the upper surface, and constructing initial average trend surfaces of the two surfaces;

[0010] acquiring prior weights according to distances of each pixel on the upper surface to the initial average trend surfaces of the two surfaces;

[0011] taking the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values of iteration, and iteratively solving a double-plane collaborative optimization objective function by using the prior weights to synchronously acquire the plane parameters of the lower surface and the plane parameters of the upper surface;

[0012] acquiring an angle of an included angle between the upper surface and the lower surface of the detector according to the plane parameters of the lower surface and the plane parameters of the upper surface.

[0013] Further, the taking the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values of iteration, and iteratively solving a double-plane collaborative optimization objective function by using the prior weights to solve the plane parameters of the lower surface and the plane parameters of the upper surface comprises:

[0014] taking the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values of iteration, and calculating all data point residuals in the current iteration;

[0015] calculating Huber dynamic weights according to an absolute median deviation of the pixel residuals on the upper surface and a Huber threshold value;

[0016] fusing the prior weights and the Huber dynamic weights to acquire pixel dynamic fusion weights;

[0017] acquiring the double-plane collaborative optimization objective function according to the pixel dynamic fusion weights, a preset data type weight, and the residuals of all data points;

[0018] iteratively solving the double-plane collaborative optimization objective function, and updating the plane parameters of the lower surface and the plane parameters of the upper surface according to a solving result.

[0019] Further, the updating the plane parameters of the lower surface and the plane parameters of the upper surface according to the solving result comprises:

[0020] acquiring updated lower surface normal vectors and upper surface normal vectors and a normal vector change amount of the last iteration;

[0021] When the lower surface normal vector change amount and the upper surface normal vector change amount are both less than a preset tolerance, it is determined that iteration converges, and the current lower surface plane parameters and upper surface plane parameters are output.

[0022] Further, the double-plane collaborative optimization objective function is: residual weighted sum minimization of the lower surface positioning point, the upper surface reference point and the upper surface pixel point cloud.

[0023] Further, the fusion of the prior weight and the Huber dynamic weight to obtain a pixel dynamic fusion weight comprises:

[0024] The prior weight of each pixel is multiplied by the corresponding Huber dynamic weight to obtain an initial fusion weight;

[0025] The initial fusion weight is normalized to obtain the pixel dynamic fusion weight for this iteration.

[0026] Further, the calculation of the Huber dynamic weight according to the residual of the upper surface pixel and the Huber threshold value comprises:

[0027] Adaptive calculation of the Huber threshold value based on the statistical distribution of all upper surface pixel residuals of the current iteration;

[0028] According to the Huber threshold value and the residual size of each pixel, the corresponding Huber dynamic weight is calculated in a piecewise function manner.

[0029] Further, the double-plane collaborative optimization objective function comprises:

[0030] The weight coefficients of the lower surface positioning point, the upper surface reference point and the upper surface pixel point cloud are different in data type distribution; wherein the weight coefficient of the lower surface positioning point is the largest, the weight coefficient of the upper surface pixel point cloud is the second, and the weight coefficient of the upper surface reference point is the smallest.

[0031] Further, the removal of duplicate data and coordinate calibration of the coordinates of the reference points of the upper surface and all pixel point clouds of the upper surface comprises:

[0032] The neighborhood deduplication method based on x-y coordinate double threshold is used for duplicate data removal processing;

[0033] According to the positioning point of the lower surface as a global reference, the data of the reference points of the upper surface and all pixel point clouds of the upper surface are subjected to coordinate calibration.

[0034] Further, the construction of the double-surface initial average trend surface comprises:

[0035] Take the corresponding coefficients of the lower surface initial plane parameters and the upper surface initial plane parameters as arithmetic mean values to obtain the plane equation coefficients of the double-surface initial average trend surface.

[0036] According to the specific embodiments of the present application, the present application also provides a computing system for detecting the parallelism of a double-surface joint optimization, comprising:

[0037] A data acquisition module is configured to acquire three-dimensional coordinates of at least four positioning points on a lower surface of a detector, three-dimensional coordinates of at least three reference points on an upper surface of the detector, and three-dimensional coordinates of all pixel point clouds on the upper surface of the detector.

[0038] A preprocessing module is configured to remove duplicate data and calibrate coordinates for the three-dimensional coordinates of the at least three reference points on the upper surface and the three-dimensional coordinates of all pixel point clouds on the upper surface.

[0039] A trend surface construction module is configured to calculate initial plane parameters of a lower surface and initial plane parameters of an upper surface, respectively, according to the positioning points on the lower surface and the reference points on the upper surface, and to construct an initial average trend surface of the double surfaces.

[0040] A prior weight generation module is configured to acquire prior weights reflecting the physical credibility of pixels according to the distances of the pixels on the upper surface to the initial average trend surface of the double surfaces.

[0041] A collaborative fitting module is configured to take the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values for iteration, to iteratively solve a double-surface joint optimization objective function by using the prior weights, and to solve plane parameters of the lower surface and plane parameters of the upper surface.

[0042] A parallelism calculation module is configured to calculate the parallelism of the upper and lower surfaces of the detector according to the plane parameters of the lower surface and the plane parameters of the upper surface.

[0043] Compared with the prior art, the above-mentioned scheme of the present application has at least the following beneficial effects:

[0044] 1. The present application provides a computing method and system for the parallelism of a double-surface joint optimization of a detector. The technical scheme of the present application constructs prior weights based on an initial average trend surface of double surfaces and fuses them with Huber dynamic weights based on iterative residuals. Through a double-suppression mechanism combining static physical credibility and dynamic fitting deviation evaluation, the influence of outliers can be more accurately and comprehensively identified and weakened. The defects of fixed weights in traditional weighted least squares and the lack of prior information in single Huber fitting are overcome, so that a more accurate plane fitting result is obtained.

[0045] 2. The application provides a kind of detector double plane joint optimization parallelism calculation method and its system, the technical scheme of the application is different from the method of step-by-step independent fitting upper and lower surfaces, adopt unified double plane collaborative optimization objective function, in one iteration process, the optimal upper and lower surface plane parameters are solved simultaneously.This collaborative method eliminates the risk of fitting error in step-by-step fitting lower surface being passed to upper surface through coordinate calibration, ensures the internal mathematical correlation between double surface parameters, reduces the overall calculation error of parallelism from the model level.

[0046] 3. The application provides a kind of detector double plane joint optimization parallelism calculation method and its system, the threshold of Huber dynamic weight of the technical scheme of the application is not fixed value, but adaptive calculation based on the statistical distribution of each iteration residual, can adapt to the error fluctuation of detector pixel point cloud in different fitting stages, ensure the scientificity and real-time of abnormal value judgment standard, further improve the adaptability and reliability of method under different measurement environment and detector model. BRIEF DESCRIPTION OF DRAWINGS

[0047] The drawings herein are incorporated into the specification and form part of the specification, show the embodiments consistent with the application, and together with the specification for explaining the principles of the application. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained from these drawings without creative labor for those skilled in the art. In the drawings:

[0048] Figure 1 A flowchart of a kind of detector double plane joint optimization parallelism calculation method shown in the embodiment of the application. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical scheme and advantages of the application more clear, the application will be further described in detail below in conjunction with the drawings, obviously, the described embodiments are only part of the embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0050] The terms used in the embodiments of the application are only for the purpose of describing specific embodiments, and are not intended to limit the application. The singular forms "a", "said" and "the" used in the embodiments of the application and the appended claims are also intended to include plural forms, unless the context clearly indicates otherwise. "Multiple" generally includes at least two.

[0051] It should be understood that the term "and / or" as used herein merely describes an associated relationship, which means that there can be three relationships, for example, A and / or B, which can represent three cases: A exists alone, A and B exist together, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0052] It should be understood that although the terms first, second, third, etc. can be used in the embodiments of the present application to describe, these descriptions should not be limited to these terms. These terms are only used to distinguish the description. For example, without departing from the scope of the embodiments of the present application, the first can also be called the second, and similarly, the second can also be called the first.

[0053] It should also be noted that the term "comprise", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the product or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such product or device. Without more limitations, the element defined by the statement "comprises a" does not exclude the presence of another identical element in the product or device comprising the element.

[0054] The optional embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0055] As shown in the figure, a method for calculating the parallelism of a detector double-plane joint optimization, comprising the following steps: Figure 1

[0056] S1, acquiring at least four positioning point coordinates of the lower surface of the detector, at least three reference point coordinates of the upper surface, and the coordinates of all pixel point clouds of the upper surface.

[0057] The embodiments of the present application collect the 4 positioning point coordinates of the lower surface of the detector, the 3 reference point coordinates of the upper surface, and the coordinate data of all 9564 pixel point clouds of the upper surface through the two-dimensional plane scanner, the comparison instrument rack and the three measurement probes in the prior art.

[0058] ​In the technical solution of this application embodiment, the three reference points on the upper surface are the minimum and sufficient conditions for defining the coarse plane of the upper surface; the fourth point among the four positioning points on the lower surface is used to verify whether the assumption of the plane of the lower surface is valid; if the four points are perfectly coplanar, then the plane fitted by the three points will naturally pass precisely through the fourth point. If the fourth point deviates significantly from the plane determined by the first three points, it indicates that there is a non-negligible deformation on the lower surface, or that there is a significant error at a certain measurement point. The number of positioning points and reference points can be increased according to the actual situation. Increasing the number of positioning points and reference points will lead to increased measurement time cost, increased measurement complexity, or the introduction of new errors; however, it can make the collected data smoother and improve resolution. Therefore, this application does not limit this, and the number of positioning points and reference points can be set according to specific circumstances.

[0059] Coordinates of the four positioning points on the lower surface: , .in, Indicates the lower surface of the detector The three-dimensional coordinates of each positioning point; Indicates the lower surface number Each positioning point Axis coordinates; Indicates the lower surface number Each positioning point Axis coordinates; Indicates the lower surface number Each positioning point Axis coordinates.

[0060] Coordinates of the three reference points on the upper surface: , ( ).in, Indicates the upper surface of the detector The three-dimensional coordinates of each reference point; Indicates the first of the upper surface benchmarks Axis coordinates; Indicates the first of the upper surface benchmarks Axis coordinates; Indicates the first of the upper surface benchmarks Axis coordinates. Data for all pixel point clouds on the upper surface: ). Indicates the upper surface of the detector The three-dimensional coordinates of a pixel point cloud; , , These represent the first and second layers of the upper surface of the detector, respectively. The x-axis, y-axis, and z-axis coordinates of a pixel point cloud.

[0061] S2, coordinates of all pixel point clouds of the upper surface are sequentially subjected to removing repeated data and coordinate calibration.

[0062] The embodiment of the application provides a preferred technical scheme, which adopts a neighborhood de-duplication method based on x-y coordinate double thresholds to remove repeated data of the data. The 9564 pixel point clouds of the upper surface are sorted in ascending order according to the x-axis coordinate and the y-axis coordinate, so that the adjacent pixel points are continuously distributed on the x-y plane, and the x-y coordinate double thresholds are both set to 5 μm. Taking the first point as a reference, the pixel points sorted subsequently are traversed, and the x-axis coordinate difference and the y-axis coordinate difference between the subsequent points and the reference point are sequentially calculated. If both the differences are less than the corresponding threshold, it is determined that the points are repeated points of the same pixel, and the current point is discarded. For example, the Δx of a certain point and the reference point is 3 μm, and the Δy is 4 μm, both of which are less than 5 μm, so it is determined that the points are repeated points, which are discarded. Wherein, Δx and Δy represent the x-axis coordinate difference and the y-axis coordinate difference respectively.

[0063] The coordinate system established by the four positioning points of the lower surface is taken as a global reference, a rigid body transformation matrix between the data point set of the three reference points of the upper surface and the coordinate system of the lower surface is calculated, the rigid body transformation matrix is applied to unify the coordinate systems of the reference points and the pixel point clouds of the upper surface through rigid body transformation, the system errors of translation and rotation caused by the adjustment of the measuring and collecting equipment are eliminated, and the coordinate calibration is completed.

[0064] S3, the initial plane parameters of the lower surface and the initial plane parameters of the upper surface are respectively calculated according to the positioning points of the lower surface and the reference points of the upper surface, and the initial average trend surfaces of the double surfaces are constructed.

[0065] Any three non-collinear points of the four positioning points of the lower surface are selected, and the initial plane parameters of the lower surface are obtained by using the least square method fitting: . Wherein, represents the initial plane parameters of the lower surface; , represents the coefficients of the x term and the y term in the initial plane equation of the lower surface respectively; represents the constant term in the initial plane equation of the lower surface. In the embodiment of the application, the initial normal vector of the corresponding initial plane parameters of the lower surface is: . Wherein, represents the initial normal vector of the initial plane parameters of the lower surface; , represents the x-axis direction coordinate and the y-axis direction coordinate of the initial normal vector of the initial plane parameters of the lower surface respectively.

[0066] According to the three reference points of the upper surface, the initial plane parameters of the upper surface are obtained by using the least square method fitting: . Wherein, represents the initial plane parameter of the upper surface; 、 respectively represent the coefficients of the x term and the y term in the initial plane equation of the upper surface; represents the constant term in the initial plane equation of the upper surface. In the embodiment of the application, the initial normal vector corresponding to the initial plane parameter of the upper surface is: . Wherein, represents the initial normal vector of the initial plane parameter of the upper surface; respectively represent the x-axis and y-axis direction coordinates of the initial normal vector of the initial plane parameter of the upper surface.

[0067] The embodiment of the application provides a preferred technical solution, calculating the average value of the initial plane parameter of the lower surface and the initial plane parameter of the upper surface (P0, P1) , , ), constructing a double-surface initial average trend surface equation , and taking the double-surface initial average trend surface equation as the calculation basis of the prior weight. Wherein, 、 、 respectively represent the coefficients of the x term, the y term and the constant term in the double-surface initial average trend surface equation.

[0068] The technical solution of the embodiment of the application provides a unified and objective physical basis for the acquisition of the prior weight in the subsequent technical solution by constructing the double-surface initial average trend surface. The double-surface initial average trend surface uses the average value of the initial plane parameter of the lower surface and the initial plane parameter of the upper surface, which is a selection based on the coordination of the double surfaces, error robustness and physical characteristics. First, the average value can reflect the coordinated and flat trend of the double surfaces, which adapts to the core requirement of parallelism calculation. The technical solution of the embodiment of the application is to solve the relative parallelism of the double surfaces, and the essence is to evaluate the relative position relationship of the double surfaces, rather than the absolute form of a single surface. The plane parameter of the double-surface initial average trend surface balances the initial inclination characteristics of the double surfaces, and the direction of the normal vector thereof is close to the average direction of the normal vectors of the double surfaces, so as to ensure that the pixel distance calculation is not biased to any surface, so that the prior weight can objectively reflect the credibility of the pixel in the relative structure of the double surfaces, rather than the local deviation of a single surface, which is highly consistent with the core target of paying attention to the relative relationship of the double surfaces in parallelism calculation, and thus improves the accuracy and precision of parallelism measurement.

[0069] S4, according to the vertical distance of each pixel of the upper surface to the double-surface initial average trend surface, the prior weight reflecting the physical credibility of the pixel is acquired.

[0070] The vertical distance of each pixel of the upper surface to the double-surface initial average trend surface is calculated, and the expression of the vertical distance is:

[0071]

[0072] in, This represents the vertical distance from the k-th pixel on the upper surface to the initial average trend surface of the two surfaces.

[0073] This application provides a preferred technical solution for correcting vertical distance by adding minute values. ;in, To represent minute values, avoid division by zero. The corrected vertical distance is obtained using the following expression:

[0074]

[0075] in, Indicates the corrected number The vertical distance from each pixel to the coarse plane.

[0076] Prior weights are generated based on the corrected inverse vertical distance ratio and normalized to ensure that the sum of the prior weights of all pixels is 1, thereby quantifying the physical credibility of pixels.

[0077] The expression for the prior weights is:

[0078]

[0079] in, Indicates the first of the upper surface The prior weights of the pixels; m represents the index variable of the pixel point cloud on the upper surface; This represents the vertical distance from the m-th pixel on the corrected upper surface to the initial average trend surface of the two surfaces.

[0080] The technical solution of this application embodiment requires prior weights to quantify the physical reliability of pixels. Reliability is determined by the vertical distance from the pixel to a reference plane. If only a single initial plane of the lower or upper surface is used as a reference, the pixel distance calculation will be distorted due to the inherent spacing or local deviation of the two surfaces. The data type weights preset on the initial average trend surface of the two surfaces reflect the macroscopic flatness trend of the upper and lower surfaces of the detector through the initial morphological characteristics of the lower and upper surfaces. This allows the distance from the pixel to the initial average trend surface of the two surfaces to directly reflect the degree of deviation of the corresponding pixel from the overall structure of the two surfaces. This provides a reasonable basis for the quantization logic of prior weights: pixels closer to the initial average trend surface of the two surfaces have higher weights, and pixels farther from the initial average trend surface of the two surfaces have lower weights. This effectively filters out sudden noise far from the overall morphology of the two surfaces.

[0081] S5. Using the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as the initial values ​​for iteration, and utilizing the prior weights, iteratively solve the dual-plane collaborative optimization objective function, and simultaneously obtain the plane parameters of the lower surface and the plane parameters of the upper surface.

[0082] S501, taking the lower surface initial plane parameter and the upper surface initial plane parameter as iteration initial values, calculating all data point residuals in the current iteration.

[0083] All data points in the embodiments of the present application include: lower surface positioning points, upper surface reference points and upper surface pixels.

[0084] The lower surface positioning point residual is expressed as: . Wherein, represents the lower surface positioning point residual.

[0085] The upper surface reference point residual is expressed as: . Wherein, represents the upper surface reference point residual.

[0086] The upper surface pixel residual is expressed as: . Wherein, represents the upper surface pixel residual.

[0087] S502, calculating the Huber dynamic weight according to the absolute median difference of the upper surface pixel residual.

[0088] Based on the statistical distribution of all upper surface pixel residuals in the current iteration, the Huber threshold is adaptively calculated.

[0089] According to the Huber threshold and the residual size of each pixel, the corresponding Huber dynamic weight is calculated in a piecewise function manner.

[0090] Based on the real-time error distribution characteristics of the upper surface pixel point cloud, the absolute values of all pixel residuals in the current iteration are extracted by using the iterative adaptive statistical calculation method, the absolute values are sorted and the median is taken, the absolute median difference of the upper surface pixel residual is calculated, and the expression of the absolute median difference is:

[0091] MAD = 1.4826 x

[0092] The expression of the Huber threshold is:

[0093] = 1.345 x MAD

[0094] Wherein, MAD represents the absolute median difference of the upper surface pixel residual; represents the median; 1.4826 represents a normal distribution correction coefficient; represents the Huber threshold of the i-th iteration.

[0095] ​The normal distribution correction coefficient makes the absolute median deviation of the upper surface pixel residual equivalent to the standard deviation σ. The selection basis of the normal distribution correction coefficient is that in the normal distribution error field, the Huber threshold value can cover about 95% of the normal residual, and only the outlier beyond the range starts the linear loss weight adjustment, realizes the double goals of retaining high precision characteristics of normal points and suppressing interference of abnormal points, and the above process needs to be re-executed after each iteration to generate the Huber threshold value corresponding to the new iteration , adapt to the error fluctuation of the pixel point cloud in different fitting stages, and determine the real-time error distribution characteristics of the upper surface pixel point cloud of the detector.

[0096] According to the Huber threshold value and the residual size of each pixel, the corresponding Huber dynamic weight is calculated in a piecewise function manner, and the expression of the Huber dynamic weight is:

[0097]

[0098] , wherein, represents the Huber dynamic weight.

[0099] S503, the prior weight and the Huber dynamic weight are fused to obtain a pixel dynamic fusion weight.

[0100] The prior weight of each pixel is multiplied by the corresponding Huber dynamic weight to obtain an initial fusion weight.

[0101] The expression of the initial fusion weight is:

[0102]

[0103] , wherein, represents the initial fusion weight.

[0104] The initial fusion weight is normalized to ensure that the initial fusion weight is 1, and the pixel dynamic fusion weight used for this iteration is obtained.

[0105] When the prior weight and the Huber dynamic weight are fused, the final pixel dynamic fusion weight is obtained by multiplication. The prior weight filters the outliers of abnormal physical positions from the global trend surface distance dimension, the Huber dynamic weight filters the outliers of abnormal fitting deviation from the current iteration residual dimension, and the weight of the outliers is reduced twice. At the same time, the prior weight can anchor the global trend of the double surfaces, avoid the deviation of the fitting from the overall shape of the detector, and the Huber dynamic weight can optimize the local detail deviation and correct the local error of the trend surface. The pixel dynamic fusion weight of the two can realize the unity of the global trend and the local detail.

[0106] ​S504, solving the double-plane collaborative optimization objective function according to the pixel dynamic fusion weight, the preset data type weight and the residual error of all data points.

[0107] In the embodiment of the application, the preset data type weight is ω1=0.4, ω2=0.1 and ω3=0.5.

[0108] wherein ω1 represents the weight coefficient of the upper surface pixel point cloud; ω2 represents the weight coefficient of the upper surface reference point; and ω3 represents the weight coefficient of the lower surface positioning point.

[0109] The application also provides a preferred embodiment, wherein the data type weight coefficients of the lower surface positioning point, the upper surface reference point and the upper surface pixel point cloud are different. The lower surface positioning point has the highest accuracy and serves as a global reference, so the weight coefficient is the largest; the upper surface pixel point cloud has a large amount of information and contains detailed information, so the weight coefficient is the second largest; and the upper surface reference point serves as an auxiliary calibration, so the weight coefficient is the smallest.

[0110] The initial iteration number, the initial plane parameter of the upper surface and the initial plane parameter of the lower surface are set, that is, the iteration number t=0, and the initial plane parameters of the upper surface and the lower surface are set as follows: The iteration optimization and parameter updating are performed, and based on the residual error of all data points, the absolute median difference and the Huber threshold value of the current iteration, the double-plane collaborative optimization objective function is obtained, and the expression of the double-plane collaborative optimization objective function is as follows:

[0111]

[0112] wherein F represents the double-plane collaborative optimization objective function; respectively represent the lower surface plane parameter and the upper surface plane parameter to be solved; represents the accuracy of the lower surface positioning point; represents the accuracy of the upper surface reference point. In the embodiment of the application represents that the accuracy of the lower surface positioning point is uniform; represents that the upper surface reference point is a precision measurement point.

[0113] ​​​​In the technical solution of the present application, the double-plane collaborative optimization objective function can eliminate error accumulation problems and ensure the collaborative correlation of the upper and lower surface plane parameters. If the lower surface plane parameters are fitted first and then the upper surface parameters are fitted based on the upper surface data, the fitting errors of the two times are easy to be transmitted to the upper surface through coordinate calibration, and the double surface parameters lack direct correlation, which finally magnifies the parallelism calculation error; the double-plane collaborative optimization objective function includes the lower surface plane parameters and the upper surface plane parameters in the same optimization framework, and the lower surface plane parameters and the upper surface plane parameters are solved synchronously through single iteration, the fitting deviation of a single surface can be offset by the complementarity of the double surface data, error transmission across the surface is avoided, and the double surface parameters always remain consistent with the real physical form of the detector, thereby providing a high-precision parameter basis for direct parallelism solving.

[0114] Meanwhile, the double-plane collaborative optimization objective function can efficiently fuse the weights of multiple types of data and significantly improve the outlier suppression capability. The double-plane collaborative optimization objective function allocates different data type weight coefficients, highlights the reference role of the lower surface positioning points, the information advantage of the upper surface pixel point cloud, and retains the auxiliary calibration value of the upper surface reference points; and the pixel dynamic fusion weight is embedded into the double-plane collaborative optimization objective function, which can realize double suppression of outliers. The contribution of outliers to the objective function is greatly weakened due to the far distance from the initial average trend surface of the double surface and the large iteration residual, while the normal pixels can dominate the parameter optimization due to the high weight and small residual, thereby solving the defect of the single surface fitting in the traditional single surface fitting that the outliers deviate the plane parameters, and ensuring the reliability of the fitting parameters.

[0115] In addition, the double-plane collaborative optimization objective function can balance the global trend and local details and ensure the stability of fitting. The double-plane collaborative optimization objective function integrates the global reference characteristics of the lower surface positioning points, the local calibration characteristics of the upper surface reference points, and the detail representation characteristics of the pixel point cloud, which can avoid the deviation of the fitting process to the local abnormal area; and the fixed allocation of the data type weight ensures that the high-reliability data dominates the iteration direction, so that the fitting plane not only fits the macro-level trend of the upper and lower surfaces of the detector, but also accurately reflects the local details at the pixel level, thereby avoiding the distortion of the global trend or the loss of local details.

[0116] The partial derivatives of the double-plane collaborative optimization objective function with respect to and are calculated, and the partial derivatives are set to 0. A double-plane collaborative optimization objective function equation group is obtained, which is a six-element linear equation group containing six unknowns. That is, the optimization objective of finding the double-plane parameters that minimize the overall fitting error is converted into a mathematical problem of solving a deterministic linear system. The technical solution provided by the embodiments of the present application can synchronously and globally optimally determine the plane parameters of the upper and lower surfaces of the detector, thereby solving the technical problem of high-precision parallelism calculation.

[0117] The expression of the six-element linear equation is:

[0118]

[0119] The six-element linear equation is solved by Cholesky decomposition, and the lower surface plane parameters and the upper surface plane parameters are updated according to the solving result.

[0120] The updated parameters are obtained Based on the updated plane parameters, the normal vectors of the upper surface and the lower surface are calculated, the lower surface normal vector is , and the upper surface normal vector is , and the lower surface normal vector and the upper surface normal vector are normalized.

[0121] S505, the lower surface plane parameters and the upper surface plane parameters are updated according to the solving result.

[0122] The updated lower surface normal vector and the upper surface normal vector are obtained, and the normal vector change amount of the last iteration is obtained.

[0123] When the lower surface normal vector change amount and the upper surface normal vector change amount are both less than a preset tolerance, it is determined that the iteration converges, and the current lower surface plane parameters and the upper surface plane parameters are output.

[0124] According to the updated lower surface plane parameters and the upper surface plane parameters, the normal vector change amount of the last iteration is calculated. That is, the length change amount of the normal vector of the t+1th iteration and the normal vector of the tth iteration is calculated, and if the change amount is less than a preset tolerance 10 -8 , it is determined that the iteration converges, the parameter updating is stopped, and the current lower surface plane parameters and the upper surface plane parameters are output. That is, when the change amount of the plane normal vector is small enough, it is indicated that the fitted plane is close to the real physical shape of the upper surface of the detector, and the improvement of the plane precision by continuing iteration is negligible, which meets the requirement of fitting stability for precise optical measurement.

[0125] In actual application, a preset auxiliary stop condition can also be used, and when the preset maximum iteration number is reached, the iteration is avoided from being trapped in an infinite loop. For example, the preset maximum iteration number is 30 times as an auxiliary stop condition. If the iteration number t reaches the maximum value, the core convergence condition that the normal vector change amount is less than the preset tolerance is still not met, and the iteration is forced to stop to ensure the calculation efficiency. Other numerical maximum iteration numbers can also be preset according to actual conditions, which are not limited in the application.

[0126] S6, according to the lower surface plane parameters and the upper surface plane parameters, the angle of the included angle between the upper surface and the lower surface of the detector is obtained. ​

[0127] The expression of the angle of the upper surface and the lower surface of the detector is:

[0128]

[0129]

[0130] wherein, represents the angle of the upper surface and the lower surface of the detector; represents the radian form of the angle of the upper surface and the lower surface of the detector; represents the finally obtained lower surface normal vector; represents the finally obtained upper surface normal vector.

[0131] The expression of the unit conversion to angle seconds is:

[0132]

[0133]

[0134] wherein, represents the angle form of the angle of the upper surface and the lower surface of the detector; represents the angle second form of the angle of the upper surface and the lower surface of the detector.

[0135] The technical scheme of the embodiment of the present application provides a more reliable physical reference for weight generation through the construction of the initial average trend surface of the double surfaces; the collaborative optimization framework ensures the synchronous optimization of the double surface parameters, avoiding error accumulation; the reasonable allocation of the data type weight, the mutual support and cooperation between the various technical schemes, and the calculation method of the double plane joint optimization significantly improve the accuracy of the parallelism calculation method, solving the error accumulation problem caused by the step-by-step fitting in the prior art.

[0136] The technical scheme of the embodiment of the present application filters abnormal outliers from the physical position dimension based on the prior weight and the initial average trend surface of the double surfaces, filters abnormal outliers from the fitting deviation dimension based on the Huber dynamic weight, realizes the full coverage and suppression of the two types of outliers through the double weight fusion, solves the defect that a single weight cannot accurately identify implicit outliers, makes the outlier suppression more accurate. The prior weight can anchor the global trend of the double surface, avoiding the deviation of the fitting from the overall shape of the detector, and the Huber dynamic weight can optimize the local detail deviation and correct the local error of the trend surface, so that the fitting plane not only fits the global flatness, but also retains the local detail characteristics, and the fitting accuracy is higher. Through the collaborative fitting model of the double planes, the upper and lower surface parameters are optimized synchronously, the normal vector correlation is directly established to solve the parallelism, and the accuracy requirement of precision measurement and the efficiency requirement of engineering application are considered, and the calculation efficiency is higher.

[0137] The application also provides a system embodiment consistent with the above-mentioned embodiment, for realizing the method steps as described in the above embodiment, based on the same name meaning explanation and the same as the above embodiment, with the same technical effect as the above embodiment, which will not be repeated here.

[0138] A system for detecting parallelism of double planes of a detector, comprising:

[0139] A data acquisition module is configured to acquire three-dimensional coordinates of at least four positioning points on a lower surface of the detector, three-dimensional coordinates of at least three reference points on an upper surface of the detector, and three-dimensional coordinates of all pixel points on the upper surface of the detector.

[0140] A preprocessing module is configured to remove duplicate data and calibrate coordinates of the pixel point cloud on the upper surface.

[0141] A trend surface construction module is configured to calculate initial plane parameters of the lower surface and initial plane parameters of the upper surface, respectively, according to the positioning points on the lower surface and the reference points on the upper surface, and to construct an initial average trend surface of the double surfaces.

[0142] A prior weight generation module is configured to acquire prior weights reflecting physical credibility of the pixel points according to distances of the pixel points on the upper surface to the initial average trend surface of the double surfaces.

[0143] A collaborative fitting module is configured to take the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values of iteration, to iteratively solve a double plane collaborative optimization objective function by using the prior weights, and to solve the plane parameters of the lower surface and the plane parameters of the upper surface.

[0144] A parallelism calculation module is configured to calculate parallelism of the upper and lower surfaces of the detector according to the plane parameters of the lower surface and the plane parameters of the upper surface.

[0145] The embodiment of the application verifies the effectiveness of the technical solution of the application through a comparative experiment, and specific data are shown in Table 1.

[0146] Table 1: Comparison of experimental results

[0147]

[0148]

[0149] It can be seen from the data of 12 groups of comparative experiments that, in the case of no outliers, the technical solution of the application can also maintain very high reference accuracy; as the proportion of outliers increases, the advantage of the technical solution of the application becomes more and more obvious. When the proportion of outliers is 6%, under 0.06 μm noise, the parallelism error calculated by the technical solution of the application is 0.166 angular seconds, which is much lower than 0.335 angular seconds of the weighted least squares and 0.522 angular seconds of the Huber method, and the accuracy is improved by more than 50%. When the proportion of outliers increases to 12% and 18%, under a variety of noise combinations, the parallelism error calculated by the technical solution of the application is always the smallest or close to the smallest among the three methods.

[0150] The RMSE of the technical solution of the application is slightly higher than that of the conventional method, which shows that the technical solution of the application gives priority to guaranteeing the accuracy of the final target parallelism, i.e. the angle between normal vectors, rather than simply pursuing the surface fitting degree RMSE of all data points. The technical problem of blindly pursuing low RMSE in the prior art, which leads to the fitting plane being deviated by outliers, thereby sacrificing the parallelism accuracy, is solved.

[0151] Finally, it should be explained that: the embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts of each embodiment can be referred to. For the system or device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0152] The above embodiments are only used to illustrate the technical solutions of the application, but not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application.

Claims

1. A method for calculating parallelism of a detector dual-plane joint optimization, characterized in that, The method comprises the following steps: obtaining coordinates of at least four positioning points on the lower surface of the detector, coordinates of at least three reference points on the upper surface of the detector, and coordinates of all pixel point clouds on the upper surface of the detector; performing duplicate data removal and coordinate calibration on the coordinates of the reference points on the upper surface and the coordinates of all pixel point clouds on the upper surface; calculating initial plane parameters of the lower surface and initial plane parameters of the upper surface respectively according to the positioning points on the lower surface and the reference points on the upper surface, and constructing initial average trend surfaces of the two surfaces; obtaining prior weights according to distances of each pixel on the upper surface to the initial average trend surfaces of the two surfaces; taking the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values of iteration, and iteratively solving a double-plane collaborative optimization objective function by using the prior weights to synchronously obtain the plane parameters of the lower surface and the plane parameters of the upper surface; obtaining an angle of the included angle between the upper surface and the lower surface of the detector according to the plane parameters of the lower surface and the plane parameters of the upper surface.

2. The computational method of claim 1, wherein, The method of taking the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values of iteration, and iteratively solving the double-plane collaborative optimization objective function to solve the plane parameters of the lower surface and the plane parameters of the upper surface comprises: taking the initial plane parameters of the lower surface and the initial plane parameters of the upper surface as initial values of iteration, and calculating residual errors of all data points in the current iteration; calculating Huber dynamic weights according to an absolute median deviation of the residual errors of the pixels on the upper surface and a Huber threshold value; fusing the prior weights and the Huber dynamic weights to obtain pixel dynamic fusion weights; obtaining the double-plane collaborative optimization objective function according to the pixel dynamic fusion weights, a preset data type weight and the residual errors of all data points; iteratively solving the double-plane collaborative optimization objective function, and updating the plane parameters of the lower surface and the plane parameters of the upper surface according to a solving result.

3. The computational method of claim 2, wherein, The method of updating the plane parameters of the lower surface and the plane parameters of the upper surface according to the solving result comprises: obtaining a normal vector change amount of the lower surface and a normal vector change amount of the upper surface from the last iteration; when the normal vector change amount of the lower surface and the normal vector change amount of the upper surface are both less than a preset tolerance, determining that the iteration converges, and outputting the current plane parameters of the lower surface and the plane parameters of the upper surface.

4. The computational method of claim 2, wherein, The double-plane collaborative optimization objective function is a weighted sum of residual errors of the positioning points on the lower surface, the reference points on the upper surface and the pixel point clouds on the upper surface.

5. The computational method of claim 2, wherein, The method of fusing the prior weights and the Huber dynamic weights to obtain the pixel dynamic fusion weights comprises: performing multiplication operation on the prior weight of each pixel and the corresponding Huber dynamic weight to obtain initial fusion weights; performing normalization processing on the initial fusion weights to obtain the pixel dynamic fusion weights used in the current iteration.

6. The computational method of claim 2, wherein, The method of calculating the Huber dynamic weights according to the residual errors of the pixels on the upper surface and the Huber threshold value comprises: adaptively calculating the Huber threshold value based on a statistical distribution of all upper surface pixel residual errors in the current iteration; calculating the corresponding Huber dynamic weight in a piecewise function manner according to the Huber threshold value and the residual error size of each pixel.

7. The computational method of claim 2, wherein, The double-plane collaborative optimization objective function comprises: The weight coefficients of the lower surface positioning points, the upper surface reference points and the upper surface pixel point cloud distribution data are different; wherein the weight coefficient of the lower surface positioning points is the largest, the weight coefficient of the upper surface pixel point cloud is the second, and the weight coefficient of the upper surface reference point is the smallest.

8. The computational method of claim 1, wherein, The coordinates of the reference points of the upper surface and all pixel point clouds of the upper surface are removed from repeated data and calibrated, comprising: The neighborhood de-duplication method based on x-y coordinate double threshold is adopted to remove repeated data processing; The coordinates of the reference points of the upper surface and all pixel point clouds of the upper surface are calibrated according to the positioning points of the lower surface as the global reference.

9. The computational method of claim 1, wherein, The double-surface initial average trend surface is constructed, comprising: The corresponding coefficients of the lower surface initial plane parameters and the upper surface initial plane parameters are taken as the arithmetic mean to obtain the plane equation coefficients of the double-surface initial average trend surface.

10. A system for implementing the parallelism of the dual-plane joint optimization of a probe according to the calculation method of any one of claims 1-9, characterized in that, Comprise: The data acquisition module is used to obtain the three-dimensional coordinates of at least four positioning points of the lower surface of the detector, the three-dimensional coordinates of at least three reference points of the upper surface, and the three-dimensional coordinates of all pixel point clouds of the upper surface; The preprocessing module is used to remove repeated data and calibrate coordinates of the three-dimensional coordinates of at least three reference points of the upper surface and the three-dimensional coordinates of all pixel point clouds of the upper surface; The trend surface construction module is used to calculate the lower surface initial plane parameters and the upper surface initial plane parameters according to the lower surface positioning points and the upper surface reference points, and to construct the double-surface initial average trend surface; The prior weight generation module is used to obtain the prior weight reflecting the physical credibility of the pixel according to the distance of each pixel of the upper surface to the double-surface initial average trend surface; The collaborative fitting module is used to take the lower surface initial plane parameters and the upper surface initial plane parameters as the iterative initial values, and to iteratively solve the double-plane collaborative optimization objective function by using the prior weight, to solve the lower surface plane parameters and the upper surface plane parameters; The parallelism calculation module is used to calculate the parallelism of the upper and lower surfaces of the detector according to the lower surface plane parameters and the upper surface plane parameters.

Images