A method to eliminate the influence of accumulated Euler angles during the detection process of white light interferometer

Abstract

The present invention discloses a method for eliminating the influence of accumulated Euler angles of a white light interferometer during the detection process, and relates to the field of ultra-precision machining detection technology. The method comprises: collecting point cloud data of the 3D surface of a microstructure sample; analyzing the characteristics of the point cloud data to determine its 3D surface shape; determining its three-dimensional structural model according to the 3D surface shape, and using an analytical method to calculate the deflection Euler angles of the point cloud data and the X, Y, and Z axes; negating the deflection Euler angles, and rotating the point cloud data space back to data that is parallel or aligned with the coordinate plane. The present invention, combined with the automatic focusing function of the white light interferometer, can more accurately locate the feature position and collect complete and effective point cloud data; eliminate the influence of errors in each acquisition process; eliminate the slight changes caused by the hardware system during long-term operation, and provide stable and reliable test results for long-term automatic measurement; significantly improve the reliability of automatic measurement detection data, and reduce the cost of manual measurement and calibration.

Description

Technical Field

[0001] The present invention belongs to the technical field of ultra-precision machining detection, and in particular relates to a method for eliminating the influence of accumulated Euler angles during the detection process of a white light interferometer. Background Art

[0002] Micro- and nanostructured materials and devices have widespread applications in micro- and nano-optoelectronics, semiconductors, micromechanics, biochips, and micro- and nano-biological detection. The fabrication and testing of micro- and nanostructures is a hot research topic. The dimensional and surface accuracy of micro- and nanostructured products are important indicators for evaluating product quality. White-light interferometers are widely used to measure microstructures. Traditional white-light interferometers provide relatively accurate single-shot manual measurements. With mass production, there is a new demand for automated dimensional and surface measurement to improve inspection efficiency. However, long-term operation of automated control equipment will generate systematic and torsional errors. Systematic errors are relatively random, often manifesting as salt-and-pepper noise, which can be removed using median filtering. However, torsional errors often accumulate over time and distance traveled, resulting in significant discrepancies between the results at the start and end of automated measurement. This accuracy is unacceptable for ultra-precision machining. In mass production, product quality stability is an essential requirement, and therefore, the results of mass inspection must also be stable and reliable.

[0003] White-light interferometers are used for automated batch inspection of microstructures. Currently, most white-light interferometers utilize an automatic measurement function, which requires the user to input the coordinates of multiple acquisition locations. However, due to factors such as long-term hardware operation and environmental changes, slight deformations inevitably accumulate between the sample stage and the lens, causing the collected sample point cloud data to deflect from the coordinate space by Euler angles. The results calculated from this deflected point cloud data are inaccurate and unstable. White-light interferometers can detect relatively common and fixed surfaces with distinct surface features. Furthermore, the optical components being inspected are ultra-precision machined products, and their point cloud data is sufficient to meet the accuracy requirements for establishing their mathematical characteristic models. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for eliminating the influence of the accumulated Euler angles during the detection process of a white light interferometer, thereby providing accurate surface data for the calculation of subsequent optical performance analysis characterization parameters by eliminating the errors caused by the Euler angles.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] The present invention provides a method for eliminating the influence of accumulated Euler angles during the detection process of a white light interferometer, including the following steps:

[0007] Collect point cloud data of the 3D surface of microstructured samples;

[0008] Analyzing features of the point cloud data, and determining a 3D surface shape of the point cloud data according to the features;

[0009] Determine its three-dimensional structure model according to the 3D surface shape, and use analytical method to calculate the deflection Euler angles α, β, and γ of the point cloud data and the X, Y, and Z axes;

[0010] Calculate the rotation matrix around the X, Y, and Z axes according to the deflection Euler angles α, β, and γ;

[0011] The point cloud data is multiplied by the rotation matrix to obtain new point cloud data that is parallel to or aligned with the X, Y, and Z axes.

[0012] As an optimal technical solution of the present invention, the 3D surface type includes an axially symmetrical surface type and a Y-direction vertically stretched surface type; the axially symmetrical surface type includes a quadrangular pyramid surface, a triangular pyramid surface, a conical surface and an aspherical surface; the Y-direction vertically stretched surface type includes a cylindrical surface and a V-surface.

[0013] As a preferred technical solution of the present invention, the three coordinate planes or characteristic planes of the 3D surface are analyzed to determine the surface of the point cloud data to be analyzed.

[0014] As a preferred technical solution of the present invention, the method for analyzing the 3D surface shape includes the following steps:

[0015] S1, select the cross-section of the XOY plane and perform edge detection;

[0016] S2, determining the type of the surface shape of the point cloud data according to the edge graph obtained by edge detection in step S1;

[0017] The determination of the point cloud data surface type specifically includes the following three situations:

[0018] If the edge graphic is a triangle, the point cloud data surface is represented as a triangular pyramid surface;

[0019] If the edge figure is a circle, the point cloud data surface type is represented as a conical surface or an aspherical surface; continue to select the cross-section of the XOZ surface and perform edge detection. If the edge figure is a triangle, the point cloud data surface type is represented as a conical surface; if the edge figure is a curve or an arc, the point cloud data surface type is represented as an aspherical surface;

[0020] If the edge graphic is a quadrilateral, the point cloud data surface type is represented as a tetrahedral surface, a V-surface or a cylindrical surface; continue to select the cross-sectional view of the XOZ surface and perform edge detection. If the edge graphic is a curve or an arc, the point cloud data surface type is represented as a cylindrical surface; if the edge graphic is a triangle, the point cloud data surface type is represented as a V-surface or a tetrahedral surface; continue to select the cross-sectional view of the YOZ surface and perform edge detection. If the edge graphic is a triangle, the point cloud data surface type is represented as a tetrahedral surface; if the edge graphic is a straight line, the point cloud data surface type is represented as a V-surface.

[0021] As a preferred technical solution of the present invention, the edge detection adopts a graphic recognition algorithm to identify the graphic as an arc or a broken line; the graphic recognition includes the following steps:

[0022] S11, acquiring edge data, performing a derivative operation on the edge data, and obtaining a derivative value;

[0023] S12, grouping the derivative values and filtering noise points;

[0024] S13, judging the edge graph according to the number of groups and the convergence of the derivative values of each group. The specific judgment process is as follows:

[0025] If the derivative values are divided into three groups, the derivative values in each group converge to the same value, and there are three convergent values in total, then the edge figure is a triangle;

[0026] If the derivative values are divided into two groups, and the derivative values in each group converge to the same value, and there are two convergent values in total, then the edge figure is a quadrilateral;

[0027] If the derivative value diverges and is a curve that changes within a positive and negative range, then the edge graph is a curve.

[0028] As a preferred technical solution of the present invention, the analysis method includes a calculation rotation axis vector analysis method and a coordinate plane section analysis method.

[0029] As a preferred technical solution of the present invention, the calculation rotation axis vector analysis method includes the following steps:

[0030] S1, calculating the rotation axis space vector of the point cloud data after being deflected by the Euler angles α, β, and γ;

[0031] S2, calculating the vector difference between the rotation axis space vector and the ideal rotation axis space vector (0, 0, 1) of the three-dimensional structure model;

[0032] S3, determining the Euler angles α, β, and γ according to the vector difference.

[0033] As a preferred technical solution of the present invention, the coordinate plane cross-section analysis method calculates the deflection Euler angles α, β, and γ by extracting and analyzing 2D graphics of the cross-section of the point cloud data on three coordinate planes.

[0034] As a preferred technical solution of the present invention, the deflection Euler angles α, β, and γ are negative, and the rotation matrix is obtained as follows:

[0035]

[0036] As a preferred technical solution of the present invention, the point cloud data is in matrix format, which is converted into (x, y, z) triple point set data.

[0037] The beneficial effects of the present invention are:

[0038] (1) The present invention combines the automatic focusing function of the white light interferometer to more accurately locate the feature position and collect complete and effective point cloud data.

[0039] (2) The present invention eliminates the influence of errors in each acquisition process one by one. The elimination process is comprehensive and is not affected by the length of automatic measurement time and the influence of random environment, and is irrelevant to the cause of the error.

[0040] (3) The present invention eliminates the irresistible minor changes brought about by the hardware system during long-term operation, eliminates the rotation and tilt parts of these minor changes, and provides stable and reliable test results for long-term automatic measurement.

[0041] (4) The present invention significantly improves the reliability of automatic measurement and detection data and reduces the cost of manual measurement and manual calibration.

[0042] (5) The present invention has been applied in actual measurement and has produced good results. The present invention is implemented in the automatic measurement function module of the white light interferometer and is applied to the batch detection of all sub-unit structures in the microlens array within the range of the entire wafer. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] For better understanding and implementation, the technical solution of the present application is described in detail below with reference to the accompanying drawings.

[0044] Figure 1 A flowchart of the steps of a method for eliminating the influence of accumulated Euler angles during the detection process of a white light interferometer provided in an embodiment of the present application;

[0045] Figure 2 A flowchart of the steps of the 3D surface analysis method provided in an embodiment of the present application;

[0046] Figure 3 A flowchart of the graphic recognition steps in edge detection provided by an embodiment of the present application;

[0047] Figure 4 A flowchart of the steps of the rotation axis vector analysis method provided in an embodiment of the present application;

[0048] Figure 5 A flow chart of the steps of the coordinate plane cross-section analysis method provided in the embodiment of the present application;

[0049] Figure 6 A flowchart of the steps of the method for eliminating the influence of cumulative Euler angles for five 3D surface types provided in an embodiment of the present application;

[0050] Figure 7 Schematic diagram of the visualization structure of six common surface types measured by the white light interferometer provided in the embodiment of the present application;

[0051] Figure 8 A schematic diagram of the algorithm for calculating the rotation axis vector method for an aspheric surface provided in an embodiment of the present application;

[0052] Figure 9 A schematic diagram of a three-coordinate cross-section of a cylinder visualized in an embodiment of the present application;

[0053] Figure 10 A flowchart of a method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer provided in another embodiment of the present application;

[0054] Figure 11 An algorithm flow chart of the method for calculating the rotation axis vector provided in an embodiment of the present application;

[0055] Figure 12 This is an algorithm flow chart of the coordinate plane section analysis method provided in the embodiment of the present application. DETAILED DESCRIPTION

[0056] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, exemplary embodiments are described in detail herein, as illustrated in the accompanying drawings. In the following description, unless otherwise indicated, identical numerals in different figures represent identical or similar elements. The embodiments described in the following exemplary embodiments are not intended to represent all possible implementations consistent with the present application. Rather, they are merely examples of methods and systems consistent with certain aspects of the present application, as detailed in the appended claims.

[0057] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0058] The following describes in detail the specific implementation methods, features and effects of the present invention in conjunction with the accompanying drawings and preferred embodiments.

[0059] Example 1

[0060] like Figure 1 As shown, the embodiment of the present application provides a method for eliminating the influence of accumulated Euler angles during the detection process of a white light interferometer, comprising the following steps:

[0061] S101, collecting point cloud data of the 3D surface of the microstructure sample;

[0062] S102, analyzing features of the point cloud data and determining the 3D surface shape of the point cloud data based on the features;

[0063] S103, determining its three-dimensional structural model based on the 3D surface shape, and using an analytical method to calculate the deflection Euler angles α, β, and γ between the point cloud data and the X, Y, and Z axes;

[0064] S104, calculating the rotation matrix around the X, Y, and Z axes based on the deflection Euler angles α, β, and γ;

[0065] S105 , multiplying the point cloud data by the rotation matrix to obtain new point cloud data that is parallel to or aligned with the X, Y, and Z axes.

[0066] Specifically, a white-light interferometer hardware device is used to collect point cloud data of the 3D surface of the microstructured sample. The collected point cloud data is analyzed one by one according to the common measurement surface types of the white-light interferometer. The 3D surface shape is determined based on its characteristics, and then the three-dimensional structural model is determined. Then, based on the three-dimensional structural model of the point cloud data, an appropriate analysis method is selected to calculate the deflection Euler angles α, β, and γ of the point cloud data and the X, Y, and Z axes. After calculating the deflection Euler angles α, β, and γ, they are negated and the rotation matrix about the X, Y, and Z axes is obtained. The negated deflection Euler angles α, β, and γ are multiplied by the rotation matrix to obtain new point cloud data parallel to or aligned with the X, Y, and Z axes. This eliminates the error of deflection rotation and converts it into point cloud data that can be directly used for comparison and analysis with the ideal mathematical expression.

[0067] In an embodiment provided in the present application, the 3D surface type includes an axially symmetric surface type and a Y-direction vertically stretched surface type; the axially symmetric surface type includes: a quadrangular pyramid surface, a triangular pyramid surface, a conical surface and an aspherical surface; the Y-direction vertically stretched surface type includes: a cylindrical surface and a V-surface.

[0068] Specifically, there are seven common surface types measured by white light interferometers that require the elimination of the influence of Euler angles: spheres, tetrahedrons, triangular pyramids, conical surfaces, aspherical surfaces, cylindrical surfaces, and V-surfaces. Spherical surfaces are completely symmetrical in all directions, and the resulting Euler angles α, β, and γ have no effect on the calculation of subsequent characterization parameters of the sphere. Therefore, this paper excludes spheres and only discusses six of the above seven surface types. These six surface types are divided into two categories: one is axisymmetric surface types, which mainly include tetrahedrons, triangular pyramids, conical surfaces, and aspherical surfaces; the other is vertically stretched surface types in the Y direction, which mainly include cylindrical surfaces and V-surfaces.

[0069] In one embodiment provided in this application, three coordinate planes or characteristic planes of a 3D surface are analyzed to determine the surface type of the point cloud data to be analyzed. In the present invention, by analyzing the three coordinate planes or characteristic planes of a 3D surface, it is determined which of the six surface types the point cloud data belongs to.

[0070] like Figure 2 As shown, in one embodiment provided in this application, the method for analyzing the above 3D surface shape includes the following steps:

[0071] S201, selecting a cross-section of the XOY plane and performing edge detection;

[0072] S202, determining the type of the surface shape of the point cloud data according to the edge graph obtained by edge detection in step S201;

[0073] The determination of the point cloud data surface type specifically includes the following three situations:

[0074] If the edge graphic is a triangle, the point cloud data surface is represented as a triangular pyramid surface;

[0075] If the edge figure is a circle, the point cloud data surface type is represented as a conical surface or an aspherical surface; continue to select the cross-section of the XOZ surface and perform edge detection. If the edge figure is a triangle, the point cloud data surface type is represented as a conical surface; if the edge figure is a curve or an arc, the point cloud data surface type is represented as an aspherical surface;

[0076] If the edge graphic is a quadrilateral, the point cloud data surface type is represented as a tetrahedral surface, a V-surface or a cylindrical surface; continue to select the cross-sectional view of the XOZ surface and perform edge detection. If the edge graphic is a curve or an arc, the point cloud data surface type is represented as a cylindrical surface; if the edge graphic is a triangle, the point cloud data surface type is represented as a V-surface or a tetrahedral surface; continue to select the cross-sectional view of the YOZ surface and perform edge detection. If the edge graphic is a triangle, the point cloud data surface type is represented as a tetrahedral surface; if the edge graphic is a straight line, the point cloud data surface type is represented as a V-surface.

[0077] Specifically, in a 3D coordinate system, the XOY plane of the coordinate system is used as the bottom plane, and the XOZ and YOZ planes are used as the side planes. First, a cross-section of the XOY plane is selected for edge detection. If the edge pattern after detection is a triangle, it means that the point cloud data surface corresponding to the cross-section is a triangular pyramid.

[0078] Furthermore, a cross-sectional view of the XOY surface is selected and edge detection is performed. If the detected edge figure is a circle, the point cloud data surface type corresponding to the cross-sectional view is a conical surface or an aspherical surface. Next, a cross-sectional view of the XOZ surface is selected and edge detection is performed. If the detected edge figure is a triangle, it means that the point cloud data surface type is a conical surface among the above two surface types; if the edge figure is a curve or an arc, it means that the point cloud data surface type is an aspherical surface.

[0079] Furthermore, select the cross-sectional view of the XOY surface and perform edge detection. If the detected edge figure is a quadrilateral, the point cloud data surface type may be a tetrahedron, a V-surface, or a cylinder. Continue to select the cross-sectional view of the XOZ surface and perform edge detection. If the edge figure is a curve or an arc, the point cloud data surface type is represented as a cylinder; if the edge figure is a triangle, the point cloud data surface type is represented as a V-surface or a tetrahedron. Continue to select the cross-sectional view of the YOZ surface and perform edge detection again. If the edge figure is a triangle, the point cloud data surface type is represented as a tetrahedron; if the edge figure is a straight line, the point cloud data surface type is represented as a V-surface.

[0080] In summary, the analysis of the surface shape of point cloud data combined with spatial coordinate axes includes the following six situations:

[0081] (1) When the edge detection of the cross-section of the XOY plane is presented as a triangle, the corresponding point cloud data surface is a triangular pyramid.

[0082] (2) When the edge detection of the cross-section of the XOY plane appears as a circle and the cross-section of the XOZ plane appears as a triangle, the corresponding point cloud data surface shape is a conical surface.

[0083] (3) When the edge detection of the cross-section of the XOY plane appears as a circle, and the cross-section of the XOZ plane appears as a curve or an arc, the corresponding point cloud data surface is aspherical.

[0084] (4) When the edge detection of the cross-section of the XOY plane presents a quadrilateral, and the cross-section of the XOZ plane presents a curve or an arc, the corresponding point cloud data surface type is a cylinder.

[0085] (5) When the edge detection of the cross-section of the XOY plane is presented as a quadrilateral, the cross-section of the XOZ plane is presented as a triangle, and the cross-section of the YOZ plane is presented as a triangle, the corresponding point cloud data surface type is a quadrangular pyramid.

[0086] (6) When the edge detection of the cross-section of the XOY plane is presented as a quadrilateral, the cross-section of the XOZ plane is presented as a triangle, and the cross-section of the YOZ plane is presented as a straight line, the corresponding point cloud data surface type is a V surface.

[0087] like Figure 3 As shown, in one embodiment provided in the present application, edge detection uses a graphic recognition algorithm to identify a graphic as an arc or a broken line; the graphic recognition includes the following steps:

[0088] S301, acquiring edge data, performing a derivative operation on the edge data, and obtaining a derivative value;

[0089] S302, grouping the derivative values and filtering out noise points;

[0090] S303: Based on the number of groups and the convergence of the derivative values of each group, the edge graph is judged. The specific judgment process is as follows:

[0091] If the derivative values are divided into three groups, the derivative values in each group converge to the same value, and there are three convergent values in total, then the edge figure is a triangle;

[0092] If the derivative values are divided into two groups, and the derivative values in each group converge to the same value, and there are two convergent values in total, then the edge figure is a quadrilateral;

[0093] If the derivative value diverges and is a curve that changes within a positive and negative range, then the edge graph is a curve.

[0094] Specifically, determining the surface shape by identifying the edge patterns of each cross section only requires qualitative analysis, that is, identifying the shape as a circle or a multi-segment polyline. The specific identification method is to take the derivative of the edge data, then group the derivative values, and filter out the small amount of noise that appears in the group.

[0095] If these derivative values can be divided into three groups, and the derivative values in each group converge to the same basically identical value, then the three groups of derivative value data will converge to three different convergence values, and these three convergence values can be obtained by linear fitting of the grouped edge data, then the edge graph is a triangle;

[0096] If these derivative values can be divided into two groups, and the derivative values in each group converge to the same basically identical value, the two groups of derivative value data will converge to two convergence values respectively. These two convergence values are different, and this convergence value can be obtained by linear fitting of the grouped edge data, then the edge figure is a quadrilateral; it should be noted that because the upper and lower slopes of the quadrilateral are the same, and the left and right slopes are also the same, the derivative values can be divided into two groups and converge to two different convergence values.

[0097] If the derivative values diverge, cannot be grouped, and are a curve that changes within a certain positive and negative range, then this edge graph is a curve.

[0098] For example, when taking the derivative of edge data, because there is a lot of data, many derivative values will be generated; grouping these derivative values can filter out the noise data; if these derivative values can be divided into three groups, first of all, it can be shown that the derivative values in each group basically converge to the same value, such as the derivative values of 0.9, 1, 1.1, 0.98, and 1.01. These five derivative values basically converge to the value 1, so the convergence value of this group can be considered to be 1. The derivative values in each group are not limited to the five derivative values in the above example; secondly, because there are three groups of derivative values, there will be a total of three convergence values; because the slopes of the three sides of a triangle cannot be the same or similar, these three convergence values are different; and this convergence value can be obtained by linear fitting the grouped edge data, so the edge shape can be considered to be a triangle. Similarly, quadrilaterals and other shapes are analyzed in the same way as above and will not be described in detail here.

[0099] In an embodiment provided in the present application, the analysis method includes a calculation rotation axis vector analysis method and a coordinate plane cross-sectional diagram analysis method.

[0100] Specifically, this embodiment uses a rotation axis vector analysis method and a coordinate plane cross-section graphic analysis method to calculate the deflection Euler angle.

[0101] In one embodiment provided in the present application, the rotation axis vector analysis method includes: calculating the rotation axis space vector after the point cloud data is deflected by the Euler angles α, β, and γ; calculating the vector difference between the rotation axis space vector and the ideal rotation axis space vector (0, 0, 1) of the three-dimensional structure model; and determining the Euler angles α, β, and γ based on the vector difference.

[0102] like Figure 4As shown, the rotation axis vector analysis method specifically includes the following steps:

[0103] S401, dividing the point cloud data by Z value, and assigning the data points therein to corresponding groups according to the Z value;

[0104] Specifically: the point cloud data set consists of many Z values. The Z values of all point cloud data are divided into N parts according to the range, and each data point in the point cloud data is grouped into the corresponding i-th group according to its Z value.

[0105] S402, selecting each set of point cloud data to obtain a 2D data set, calculating the center point of the cross-sectional figure, and obtaining the set of 3D coordinate points;

[0106] Specifically: select each set of point cloud data (x i ,y i ), a new 2D data set is obtained, each set of 2D data can be regarded as this height z i The cross-sectional 2D graph below is calculated, and the center point (x c ,y c ). i The value is obtained by averaging the Z values in the point cloud dataset, merging (x c ,y c ,z i ), which is the 3D coordinate point of this group on the rotation axis.

[0107] S403, searching for all 3D coordinate points in the point cloud data to obtain the vertex V; calculating the center 3D coordinate point of all Z value groups, merging it with the vertex V to form a 3D coordinate point set, and fitting it into a rotation axis space vector;

[0108] Specifically: find the point with the largest Z value among all 3D coordinate points in the point cloud data as the vertex V of this axisymmetric figure, and at the same time calculate the 3D coordinate point of the center of the cross-sectional figure under all Z value groups, add the vertex V, and merge them to form a 3D coordinate point set. The 3D coordinate point set is formed into a spatial straight line, and it is fitted into a 3D straight line vector P(x, y, z). The 3D straight line vector P(x, y, z) is the new rotation axis space vector after the microstructure sample is deflected to generate the Euler angle.

[0109] S404, calculating the deflection Euler angles α and β according to the angles between the projection lines of the rotation axis space vector on the YOZ plane and the XOZ plane and the ideal rotation axis space vector;

[0110] Specifically, the angle between the projection line of the deflected rotation axis space vector P on the YOZ plane and the ideal rotation axis space vector (0,0,1) is the Euler angle α; the angle between the projection line of the deflected rotation axis space vector P on the XOZ plane and the ideal rotation axis space vector (0,0,1) is the deflection Euler angle β.

[0111] S405, calculating the deflection Euler angle γ;

[0112] Specifically, determine the XOY plane rotation angle γ, which is perpendicular to the rotation axis space vector. This is calculated by taking a cross-sectional shape of the XOY base. If the cross-sectional shape is a triangle or quadrilateral, use edge straight line fitting to calculate the slopes of each side. The triangle is then rotated to an isosceles triangle, and the quadrilateral to a rectangle. The rotation angle at this point is γ. If the XOY plane cross-sectional shape is a circle, set γ to 0. Since a circle is a completely symmetrical shape, the rotation angle γ has no effect on its point cloud data and does not need to be calculated.

[0113] In this embodiment, the deflection of the Euler angle α is also called the rotation angle α; the deflection of the Euler angle β is also called the rotation angle β; and the deflection of the Euler angle γ is also called the rotation angle γ.

[0114] In one embodiment provided in the present application, the coordinate plane cross-section analysis method calculates the Euler angles by extracting and analyzing 2D graphics of cross sections of point cloud data on three coordinate planes.

[0115] like Figure 5 As shown in FIG, the coordinate plane section analysis method specifically includes the following steps:

[0116] S501, extracting the cross-sectional figure of the XOY plane, calculating the slope of the oblique line, and calculating the rotation angle γ;

[0117] Specifically, extract the cross-sectional figure of the XOY plane, which should be a figure with two oblique lines as its boundary, calculate the slope of the oblique lines, and rotate them to be vertical. The rotation angle is γ.

[0118] S502, extracting a data list of the 2D data set, obtaining a deflection oblique line after data processing, calculating the slope and inclination angle of the deflection oblique line, and obtaining a rotation angle α;

[0119] Specifically, extract the data list of each (x,z) 2D data set in the stretching direction (Y axis) and parallel to the XOZ plane, find the maximum value of the Z value in each 2D series, and compare it with each y i Merge to obtain a 2D curve with the highest point along the stretching direction. This curve should be a slant line deflected at a certain angle from the OY axis. Calculate the slope and inclination angle of this deflected slant line, and rotate the deflected slant line to be parallel to the OY axis. The rotation angle at this time is α.

[0120] S503, extracting the cross-sectional graph of the XOZ plane, obtaining an inclined straight line after data processing, calculating the slope and inclination angle of the inclined straight line, and obtaining a rotation angle β;

[0121] Specifically, the cross-sectional figure of the XOZ surface is extracted. The cross-sectional figure is an arched curve or an inverted V-shaped broken line rotated at a certain angle, wherein the arched curve is a cylindrical surface and the inverted V-shaped broken line is a V-surface.

[0122] For the arch curve, the X value traverses the arch curve from left to right to the highest point of the Z value, and the current X value is x l , Z value is z i , and interpolate to get the same height z on the right i The x corresponding to the right side of the highest point of the lower arch curve r value, take x l ,x r The midpoint of i , find all (x i ,z i ) is merged into the point set L, and a straight line is fitted to the point set L to obtain an inclined straight line. The slope and inclination angle of L are obtained, and the angle required to rotate it to vertical is β.

[0123] For the inverted V-shaped broken line, a straight line is fitted to the left and right sides to obtain an inclined straight line, and their slopes are respectively calculated. The two inclined straight lines are rotated to the angle required for bilateral symmetry, and the rotation angle is β.

[0124] In one embodiment provided in the present application, the deflection Euler angles α, β, and γ are negative, and the rotation matrix is obtained as follows:

[0125]

[0126] Specifically, in the embodiment of the present application, the rotation angle α, the rotation angle β, and the rotation angle γ are negative, and the rotation matrix M (Mx, My, Mz) of the rotation around the three axes is calculated using the formula:

[0127] By multiplying the point cloud data by the rotation matrix M, the sample's point cloud data can be rotated back into 3D point cloud data aligned or parallel to the three axes or coordinate planes. In this embodiment, alignment is also referred to as perpendicularity. This eliminates errors caused by deflection and rotation, converting the data into point cloud data that can be directly compared and analyzed with the ideal mathematical expression.

[0128] In an embodiment provided in the present application, if the point cloud data is in a matrix format, it is converted into (x, y, z) triple point set data.

[0129] In summary, the present invention has the following advantages and features:

[0130] (1) Currently, most automatic measurement functions of white light interferometers require users to input the coordinates of multiple acquisition positions. However, factors such as sample stage deformation and sample warping cause random and cumulative errors between the remote calculated coordinates and the actual sampled coordinates. The method of the present invention, combined with the automatic focusing function of the white light interferometer, can ensure more accurate positioning of feature positions and collect complete and effective point cloud data.

[0131] (2) The method of the present invention eliminates the influence of errors in each acquisition process one by one. The elimination process is comprehensive and is not affected by the length of automatic measurement time and the random environment, and is irrelevant to the cause of the error.

[0132] (3) The long-term operation of the hardware system will bring about irresistible minor changes, and these small errors are unacceptable in the field of micro-nano processing. The present invention eliminates the rotation and tilt parts of these minor changes, providing stable and reliable test results for long-term automatic measurement.

[0133] (4) This method significantly improves the reliability of automated measurement data and reduces the cost of manual measurement and calibration. Furthermore, the present invention has been implemented in a white-light interferometer automated measurement module and applied to batch testing of all subunit structures in a microlens array across an entire wafer.

[0134] Example 2

[0135] The method described in this application is implemented through a computer program algorithm and is applicable to a variety of common inspection surfaces for white-light interferometers. Extensive data and experiments have confirmed that the method described in this application can accurately eliminate the deflection Euler angle between the sample stage and the lens within a relatively small deflection range.

[0136] The functions implemented in this application are integrated into the white light interferometer fully automatic detection and analysis software system. The functional module of this application runs after collecting point cloud data and before the module for analyzing and calculating optical device characterization parameters.

[0137] In the embodiments provided in this application, 3D surface types include aspheric, cylindrical, V-surface, conical, quadrangular and triangular pyramidal surfaces. Commonly used white light interferometers to detect the size of microlens array products are mainly based on the above structures.

[0138] like Figure 6 As shown, for the above six 3D surface types, the embodiment of the present application provides a method for eliminating the influence of the accumulated Euler angles during the detection process of the white light interferometer, which specifically includes the following implementation steps:

[0139] S601: Use white light interferometer hardware equipment to collect 3D point cloud data of microstructure samples.

[0140] The point cloud data is provided in the form of a matrix. The row and column labels can be regarded as x and y values in three-dimensional space, and the corresponding value of the matrix is the z value, such as Figure 7 As shown in the figure, it is the visualization structure of six common surface types measured by white light interferometer. Figure 7 It can be seen that the original point cloud data measured by the white light interferometer has produced a considerable rotation offset and a certain angular deviation from the spatial coordinate axis.

[0141] S602: Analyze the common measurement surface types of white light interferometer one by one and determine their mathematical models.

[0142] Specifically, the data of the XOY cross-section graph is taken for edge detection, and the main algorithms used are: linear normalization and edge detection;

[0143] To identify the shape of an edge detection graphic, you only need to identify whether the shape is a circle or a polyline. The identification method is: take the derivative of the acquired edge data. If the derivative value is a curve that changes within a certain range, it is a circle; if the derivative value converges to three basically the same values, it is a triangle; if the derivative value converges to two basically the same values, it is a quadrilateral.

[0144] If the shape of the XOY surface edge detection figure is a triangle, the surface type is a triangular pyramid;

[0145] If the shape of the XOY surface edge detection figure is a circle, the surface may be a conical surface or an aspherical surface; take the cross section of the XOZ surface and identify the shape of the figure. If it is a triangle, the surface is a conical surface; if it is a circle, the surface is an aspherical surface;

[0146] If the XOY surface edge detection pattern is a quadrilateral, the surface may be a quadrangular pyramid, a V-shaped surface, or a cylinder. Take a cross section of the XOZ surface and identify the shape. If it is a circle, the surface is a cylinder; if it is a triangle, the surface is a quadrangular pyramid or a V-shaped surface. Then take a cross section of the YOZ surface and identify the shape. If it is a triangle, the surface is a quadrangular pyramid; if it is a single inclined straight line, it is a V-shaped surface; if it is a circle, it is an aspherical surface or a conical surface; if it is three straight lines, it is a triangular pyramid.

[0147] S603, calculating the Euler angles α, β, and γ according to the surface shape determined in step S602;

[0148] Specifically: the calculation of the rotation axis vector method is used for the conical surfaces, triangular pyramids, quadrangular pyramids, and aspherical surfaces of the three-dimensional axisymmetric surface type; the coordinate surface section analysis method is used for the cylindrical surfaces and V surfaces of the vertical stretching surface in the Y direction.

[0149] The specific analysis is as follows:

[0150] (1) Conical surface: Find the point (x, y, z) with the maximum z value in the point cloud data. Group the point cloud data by height. Assume that there are n groups, each with a height of △z = Z / n. The height of the i-th group is between [△z*i, △z*(i+1)]. The average z value in the (x, y, z) triple list in each group is Z0. The center (X0, Y0) of the circle is obtained by the x, y binary list of the aspheric curve fitting (ellipse fitting) triple list. Add Z0 to get the projection center (X0, Y0, Z0) at a certain height value in the group.

[0151] All projection centers and apexes are 3D points. Together, they form the 3D curve of the cone's axis of symmetry. A straight line in 3D space fits this 3D curve. The axis of symmetry space vector L is the angle between the cone and the ideal axis of symmetry space vector L0 (0, 0, 1). The angle between the projection line Ly of the axis of symmetry space vector L on the YOZ plane and the OZ vector on the YOZ plane is the Euler angle α. The angle between the projection line Lx of the axis of symmetry space vector L on the XOZ plane and the OZ vector on the XOZ plane is the Euler angle β.

[0152] Since the bottom surface of the cone is a circle and the circle is symmetrical in all directions, the generation of the Euler angle γ has no effect on the point cloud data, and the γ value of the cone surface can be 0.

[0153] (2) Triangular pyramid: Find the point (X, Y, Z) with the maximum z value in the point cloud data, group the point cloud data according to the height value, and set it to n groups. The height of each group is △z=Z / n, and the height of the i-th group is between [△z*i,△z*(i+1)]. The average z value in the list of (x, y, z) triples in each group is Z0. Take (x, y) in each group (x, y, z) to get a binary array, and get a closed triangular surface. Use the quadratic circle fitting method to fit the circle to get the center (Xc, Yc). Then traverse the binary (x, y) to get the three points farthest from the center, which are the three vertices of the triangle. Calculate the circumcircle center (X0, Y0) of the triangle through these three points. Add Z0 to get the projection center (X0, Y0, Z0) at a certain height value in the group.

[0154] All projection centers and apexes are 3D points. Together, they form the 3D curve of the triangular pyramid's axis of symmetry. A straight line in 3D space fits this 3D curve. The symmetry axis space vector L is the angle between the triangular pyramid and the ideal symmetry axis space vector L0 (0, 0, 1). The angle between the projection line Ly of the symmetry axis space vector L on the YOZ plane and the OZ vector on the YOZ plane is the Euler angle α. The angle between the projection line Lx of the symmetry axis space vector L on the XOZ plane and the OZ vector on the XOZ plane is the Euler angle β.

[0155] Take a screenshot of the XOY surface of the point cloud data. The XOY surface screenshot of the triangular pyramid is a rotated isosceles triangle. Use least squares circle fitting to find the circumcenter of the triangle. Group the three sides of the triangle by the center of the circle, find the inclination angles of the three sides, and rotate them into an isosceles triangle with the base parallel to the X-axis. The rotation angle is γ.

[0156] (3) Pyramid: Find the point (X, Y, Z) with the maximum z value in the point cloud data. Group the point cloud data according to the height value. Assume that there are n groups, and the height of each group is △z = Z / n. The height of the i-th group is between [△z*i,△z*(i+1)]. The average z value in the list of (x, y, z) triples in each group is Z0. Take (x, y) in each group (x, y, z) to get a binary array, and get a closed quadrilateral domain. Use the quadratic circle fitting method to fit the circle center (Xc, Yc). Then traverse the binary (x, y) to get the four points farthest from the center, which are the four vertices of the quadrilateral. Find the intersection point (X0, Y0) obtained by the intersection of the four points. Add Z0 to get the projection center (X0, Y0, Z0) at a certain height value in the group.

[0157] All projection centers and apexes are 3D points. Together, they form the 3D curve of the symmetry axis of the tetrahedron. A straight line in 3D space fits this 3D curve. The symmetry axis space vector L is the angle between the tetrahedron and the ideal symmetry axis space vector L0 (0, 0, 1). The angle between the projection line Ly of the symmetry axis space vector L on the YOZ plane and the OZ vector on the YOZ plane is the Euler angle α. The angle between the projection line Lx of the symmetry axis space vector L on the XOZ plane and the OZ vector on the XOZ plane is the Euler angle β.

[0158] Take a screenshot of the XOY plane of the point cloud data. The bottom of the quadrangular pyramid is a rotated rectangle. Use least squares circle fitting to find the circumcenter of the quadrilateral. Group the four edges by the circle center and calculate the inclination angles of each edge, which should be two roughly equal values. Rotate the rectangle by an angle of γ to create a rectangle with two sides parallel to the x and y axes.

[0159] (4) Aspheric surface: The algorithm for aspheric surface is almost exactly the same as that for conical surface. Figure 8 As shown, this embodiment provides a schematic diagram of the algorithm for calculating the rotation axis vector method for aspheric surfaces, where the aspheric surface is a rotation axis symmetrical figure. Figure 7 , it can be seen Figure 8 The first picture is Figure 7 From the projection image of the aspheric surface on the XOZ plane, it can be clearly seen that the bottom of the aspheric surface is not parallel to the X axis, and a certain angle of inclination is generated. This angle is the rotational deflection angle of the aspheric surface. Figure 8The second figure is a curve rotation diagram of the aspheric surface. The curve rotates spirally, resulting in the deflection angle in the first figure. Figure 8 The third figure is the projection of the second figure on the XOZ plane. It can be clearly seen that the figure at this time is more like a series of line segments, and there is a certain distance between each line.

[0160] Specifically, we search for the point (X, Y, Z) with the maximum z value in the point cloud data. We then group the point cloud data by height, assuming n groups, each with a height of △z = Z / n, and the height of group i between [△z*i, △z*(i+1)]. The average z value in the (x, y, z) triplet list within each group is Z0. Using the aspheric curve fit (ellipse fit) triplet list, we find the center (X0, Y0) of the circle using the x, y binary. This is then added to Z0 to obtain the projection center (X0, Y0, Z0) at a given height within the group.

[0161] All projection centers and apex points are 3D points. Together, they form the 3D curve of the symmetry axis of the aspheric surface. A straight line in 3D space fits this 3D curve. The symmetry axis space vector L is the angle between the aspheric surface and the ideal symmetry axis space vector L0(0,0,1). The angle between the projection line Ly of the symmetry axis space vector L on the YOZ plane and the OZ vector on the YOZ plane is the Euler angle α. The angle between the projection line Lx of the symmetry axis space vector L on the XOZ plane and the OZ vector on the XOZ plane is the Euler angle β.

[0162] Since the bottom surface of an aspheric surface is a circle and the circle is symmetrical in all directions, the generation of the Euler angle γ has no effect on the point cloud data, and the γ value of the conical surface can be 0.

[0163] (5) Cylindrical surface: Figure 9 As shown, the present embodiment provides a schematic diagram of a three-coordinate cross-section of a cylinder, combined with Figure 7 and Figure 9 , as can be seen from the figure, Figure 9 The first picture is Figure 7 The first figure is the projection of the cylinder on the YOZ plane, and the second figure is the projection of the cylinder on the XOZ plane. Combining the first and second figures, it can be seen that the original point cloud data has produced a certain angle of inclination, resulting in a rotation offset angle; Figure 9The third figure is the projection of the cylinder on the XOY plane. The bright white spot in the center is the projection of the cylinder. Specifically, the point cloud data is read row by row (in the y-direction), generating row (x, z) 2D curves. Each 2D curve should have a peak. A polynomial is fitted to this 2D curve to obtain the highest point, which corresponds to the highest point in the y-direction of each cylinder. The cylinder should be parallel to the y-direction, so the curve formed by connecting the highest points is parallel to the y-axis (angle 0). If it forms a certain angle with the y-axis, this angle represents the angle α of the point cloud data rotation around the x-axis.

[0164] Record the x and y values of the highest point, and fit the resulting (x,y) curve to a straight line. This represents the angle γ by which the point cloud data is rotated about the z-axis. Alternatively, analyze the cross-section of the XOY plane, which should be a rotated rectangle. Find its center by fitting a circle, group it into four sides, and calculate the slope and inclination angle of each side. Rotate the rectangle to a shape with sides parallel to the x and y axes. The rotation angle is γ.

[0165] The projection data of the XOZ surface should be a rotated single peak curve. Take the part of the area on both sides of the peak to analyze. Find the highest point V of the curve, and take the value of x from the leftmost (minimum) to the highest point. Take each x li The corresponding z i value, and at the same time find this z i The x corresponding to the right side of the peak ri Value, find (x li ,x ri )'s midpoint x i , and (x i ,z i ) is the symmetrical point at this height. Find all (x i ,z i ) Set L is the symmetry axis of the 2D curve. A straight line is fitted to the set L to obtain its slope and tilt angle. Rotating it to vertical is the angle β of the point cloud data rotated around the y-axis.

[0166] (6) V-plane: The number of rows of point cloud data is rows. Read the point cloud data line by line (y direction). The number of rows of point cloud data is rows. Read the point cloud data line by line (y direction). Get rows of (x, z) 2D polylines. The relative heights (highest - lowest) of these 2D polylines should be basically equal. Consider each 2D curve as the two sides of a centrally symmetrical triangle formed by two straight lines. Determine the actual highest point of the curve. Take the points on the left and right sides and fit them with straight lines to obtain the inclination angles of the two straight lines. At the same time, according to the equations of the two fitted lines, the intersection of the two straight lines is the highest point. If the left and right sides are symmetrical, it means that the V-plane is parallel to the x-axis. The difference between the angles of the left and right lines is the angle α of the point cloud data rotation around the x-axis.

[0167] Record the (x, y, z) values of each of the highest points mentioned above, fit the (x, y) curve into a straight line, and the angle of this straight line is the angle γ of the point cloud data rotating around the z-axis. Fit the (x, z) curve into a straight line, and the angle of this straight line is the angle β of the point cloud data rotating around the y-axis.

[0168] S604 , according to the deflection angles α, β, and γ obtained in step S603 , the affected 3D point cloud data in the sample is rotated back to data parallel to or aligned with the X, Y, and Z axes of the three-dimensional space.

[0169] Specifically, the deflection angles α, β, and γ obtained in step S603 are negated to obtain the rotation matrix around the X, Y, and Z axes. The rotation matrix is as follows:

[0170]

[0171] Multiply the initial point cloud data by the above three matrices in sequence to obtain new point cloud data. The new point cloud data is the point cloud data that eliminates the influence of the deflection Euler angle and is perpendicular or parallel to the X, Y, and Z coordinate planes.

[0172] like Figure 10 As shown, the 3D point cloud data of the microstructure detection sample collected by the white light interferometer hardware device is input into the system, and the three-dimensional structural model of the 3D curve is determined by the projection graphic analysis method on the X, Y, and Z three-axis coordinate planes; if the 3D surface is a rotational axis symmetric figure, the point cloud data is used to calculate the spatial symmetry axis vector, and then the α and β in the Euler angle are determined, and then the XOY plane projection graphic is used to calculate γ; if the 3D surface is a Y-direction vertical stretching figure, the coordinate plane projection graphic analysis is used to calculate α, β, and γ; the Euler angles α, β, and γ are obtained, and the Euler angles are negative and the rotation matrix is calculated to obtain the point cloud data of the parallel or perpendicular coordinate plane and output. The above-mentioned rotational axis symmetric figures include quadrangular pyramidal surfaces, triangular pyramidal surfaces, conical surfaces, and aspherical surfaces; the above-mentioned Y-direction vertical stretching figures include cylindrical surfaces and V-surfaces. In this embodiment, the above-mentioned quadrangular pyramidal surfaces are also called quadrilateral pyramidal surfaces, and the triangular pyramidal surfaces are also called triangular pyramidal surfaces; the specific calculation method and acquisition steps of the above-mentioned Euler angles can refer to the above-mentioned embodiment content and will not be repeated here.

[0173] like Figure 11As shown in the figure, the method of calculating the rotation axis vector is used to process the axisymmetric graphics, specifically: the point cloud data is grouped by height, the average z value of each group is zi, the (x, y) binary array fitting is used to find the center point (xi, yi), zi and the center point are combined to form a point data set (xi, yi, zi), and the rotation axis space vector L (x0, y0, z0) is obtained by spatial straight line fitting; the angle between the projection of L on the YOZ plane and the OZ axis is obtained, and the rotation angle α is obtained by rotating around the x-axis; the angle between the projection of L on the XOZ plane and the OZ axis is obtained, and the rotation angle β is obtained by rotating around the y-axis; the bottom cross-section graphic analysis method is used to analyze the bottom cross-section, which includes triangles, quadrilaterals and circles. The triangles and quadrilaterals are rotated until the bottom sides are parallel to OX, and the rotation angle γ is obtained by rotating around the z-axis; and the Euler angles α, β, and γ are obtained.

[0174] like Figure 12 As shown, the coordinate plane section analysis method is used to process the vertical stretching graphics in the Y direction. Specifically, the y value is traversed to the section parallel to the xoz plane. Each yi corresponds to the set L(y, z) of the highest point z value (yi, zi) of the cross-sectional surface. The slope of the set L(y, z) is obtained by straight line fitting. The angle between the fitting line and the OY axis is calculated, and the rotation angle α is obtained by rotating around the x-axis. For the XOZ surface cross-sectional graphics, if it is an inclined triangle, the slopes of the left and right sides are calculated. Its base is parallel to the OX axis, and the angle between it and the OZ axis is calculated. The rotation angle β is obtained by rotating around the x-axis. If it is an inclined elliptical arc, the vertex deflection angle is calculated, the arc is symmetrical on both sides, and the angle between it and the OZ axis is calculated. The rotation angle β is obtained by rotating around the x-axis. For the bottom cross-sectional graphics analysis, it is necessary to calculate the inclined rectangle so that its length and width are parallel to the OX axis and the OY axis respectively. The rotation angle γ is obtained by rotating around the z axis to obtain the Euler angles α, β, and γ.

[0175] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for eliminating the influence of accumulated Euler angles during the detection process of a white light interferometer, characterized by: The steps include: Collect point cloud data of the 3D surface of microstructured samples; Analyzing features of the point cloud data, and determining a 3D surface shape of the point cloud data according to the features; Determine its three-dimensional structure model according to the 3D surface shape, and use analytical method to calculate the deflection Euler angles α, β, and γ of the point cloud data and the X, Y, and Z axes; The rotation matrix around the X, Y, and Z axes is calculated based on the deflection Euler angles α, β, and γ, specifically: The rotation matrix is obtained by negating the deflection Euler angles α, β, and γ: ； The point cloud data is multiplied by the rotation matrix to obtain new point cloud data that is parallel to or aligned with the X, Y, and Z axes.

2. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 1, characterized in that: The 3D surface types include axisymmetric surface types and Y-direction vertical stretching surface types; the axisymmetric surface types include quadrangular pyramidal surfaces, triangular pyramidal surfaces, conical surfaces and aspherical surfaces; the Y-direction vertical stretching surface types include cylindrical surfaces and V-surfaces.

3. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 2, characterized in that: The three coordinate planes or characteristic planes of the 3D surface are analyzed to determine the surface of the point cloud data to be analyzed.

4. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 3, characterized in that: Method for analyzing the 3D surface The following steps are involved: S1, select the cross-section of the XOY plane and perform edge detection; S2, determining the type of the surface shape of the point cloud data according to the edge graph obtained by edge detection in step S1; The determination of the point cloud data surface type specifically includes the following three situations: If the edge graphic is a triangle, the point cloud data surface is represented as a triangular pyramid surface; If the edge figure is a circle, the point cloud data surface type is represented as a conical surface or an aspherical surface; continue to select the cross-section of the XOZ surface and perform edge detection. If the edge figure is a triangle, the point cloud data surface type is represented as a conical surface; if the edge figure is a curve or an arc, the point cloud data surface type is represented as an aspherical surface; If the edge graphic is a quadrilateral, the point cloud data surface type is represented as a tetrahedral surface, a V-surface or a cylindrical surface; continue to select the cross-sectional view of the XOZ surface and perform edge detection. If the edge graphic is a curve or an arc, the point cloud data surface type is represented as a cylindrical surface; if the edge graphic is a triangle, the point cloud data surface type is represented as a V-surface or a tetrahedral surface; continue to select the cross-sectional view of the YOZ surface and perform edge detection. If the edge graphic is a triangle, the point cloud data surface type is represented as a tetrahedral surface; if the edge graphic is a straight line, the point cloud data surface type is represented as a V-surface.

5. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 4, characterized in that: The edge detection uses a graphic recognition algorithm to identify the graphic as an arc or a broken line; the graphic recognition includes the following steps: S11, acquiring edge data, performing a derivative operation on the edge data, and obtaining a derivative value; S12, grouping the derivative values and filtering noise points; S13, judging the edge graph according to the number of groups and the convergence of the derivative values of each group. The specific judgment process is as follows: If the derivative values are divided into three groups, the derivative values in each group converge to the same value, and there are three convergent values in total, then the edge figure is a triangle; If the derivative values are divided into two groups, and the derivative values in each group converge to the same value, and there are two convergent values in total, then the edge figure is a quadrilateral; If the derivative value diverges and is a curve that changes within a positive and negative range, then the edge graph is a curve.

6. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 1, characterized in that: The analysis method includes a calculation rotation axis vector analysis method and a coordinate plane section analysis method.

7. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 6, characterized in that: The method for calculating the rotation axis vector analysis comprises the following steps: S1, calculating the rotation axis space vector of the point cloud data after being deflected by the Euler angles α, β, and γ; S2, calculating the vector difference between the rotation axis space vector and the ideal rotation axis space vector (0, 0, 1) of the three-dimensional structure model; S3, determining the Euler angles α, β, and γ according to the vector difference.

8. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 6, characterized in that: The coordinate plane cross-section analysis method calculates the deflection Euler angles α, β, and γ by extracting and analyzing 2D graphics of cross sections of the point cloud data on three coordinate planes.

9. The method for eliminating the influence of accumulated Euler angles during detection by a white light interferometer according to claim 1, characterized in that: If the point cloud data is in matrix format, it is converted into (x, y, z) triple point set data.

Images