Glass internal modification positioning method based on sd-oct and binocular vision

Abstract

The application discloses a glass internal modification positioning method based on SD-OCT and binocular vision, and comprises the following steps: performing spatial alignment and calibration on an SD-OCT system and a binocular vision system; calibrating the pixel accuracy of two directions of a two-dimensional tomographic image B-Scan of the SD-OCT system and the depth direction pixel accuracy of an A-Scan; establishing the conversion relationship between a binocular system pixel coordinate system and a world coordinate system; establishing the coordinate mapping relationship between the SD-OCT system and the binocular vision system; constructing a plane equation for the upper and lower surfaces of a glass modification feature; simultaneously shooting the three-dimensional coordinates of the modification feature end points by the binocular vision; converting the three-dimensional coordinates of the modification feature end points in the binocular vision system to the SD-OCT system, and solving the distance from the modification feature end points to the upper and lower surfaces of the glass in the SD-OCT system, so as to realize the positioning of the glass internal modification. According to the application, the accurate positioning of the modification feature can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of laser modification and positioning detection, and particularly relates to a glass internal modification positioning method based on SD-OCT and binocular vision. BACKGROUND

[0002] When the focused ultra-short infrared laser pulse acts on the glass interior, it presents a localized high intensity in time and space, and then completes the permanent modification of the glass. The Chinese patent document with the publication number CN120004515A discloses a method for femtosecond laser to process a microchannel in glass, so that the femtosecond laser is used to prepare a modified region in the glass material. The Chinese patent document with the publication number CN103025475A discloses a laser processing method, which focuses the laser in the interior of a processing object formed by glass to form a modified region.

[0003] The modified glass has differences in refractive index, absorption coefficient, chemical reaction activity, etc. from the unmodified glass. When the glass modification position deviates, it will cause a series of problems, such as causing three-dimensional storage information error in the optical storage field, causing depth deviation in induced etching, and then causing the performance of optical devices to decline or even fail.

[0004] At present, the positioning of the modified feature is mainly realized by offline sampling, that is, using a microscope to observe the cross section of the cut glass. This method destroys the sample structure and has low measurement efficiency.

[0005] The non-destructive detection methods for the internal features of the glass mainly include machine vision, millimeter wave tomography, over-focus scanning optical microscope, time-domain and frequency-domain terahertz tomography, ultrasonic detection, differential interference contrast microscope and spectral domain optical coherence tomography. However, these non-destructive detection methods have certain limitations in positioning the modified features. The machine vision can collect the morphology of the modified features in the glass, but it is difficult to shoot the upper and lower surfaces of the modified glass. The detection accuracy of the time-domain and frequency-domain terahertz tomography, ultrasonic detection and other three-dimensional scanning technologies is low, which is difficult to meet the detection accuracy of microns. The detection rate of the differential interference contrast microscope and the over-focus scanning optical microscope is low, which is difficult to meet the online non-destructive demand. In comparison, the spectral domain optical coherence tomography (SD-OCT) imaging has the advantages of high resolution, high speed non-contact, online real-time detection, etc.

[0006] However, in the detection of the modified features by using a single SD-OCT, when the probe light is perpendicular to the incident, the backscattering light of the modified features is much smaller than the reflection light of the glass surface, and it is difficult to realize the synchronous measurement of the modified features and the glass surface. When the probe light is obliquely incident, the reflection light of the glass surface cannot enter the detection arm, at this time, the upper and lower surfaces are almost impossible to image, only the backscattering light of the modified features reaches the detector. SUMMARY

[0007] The application provides a glass internal modification positioning method based on SD-OCT and binocular vision, calibrates an SD-OCT system and a binocular vision system, calculates relative spatial geometric position relations by using a calibration result, and realizes accurate positioning of a modification feature.

[0008] A glass internal modification positioning method based on SD-OCT and binocular vision comprises the following steps:

[0009] (1) calibrate the SD-OCT system and the binocular vision system in space to make the field of view ranges of the two systems overlap on a target glass region;

[0010] (2) calibrate the pixel accuracy of two directions of a two-dimensional tomographic image B-Scan of the SD-OCT system and the depth direction pixel accuracy of an A-Scan, and simultaneously calibrate the internal and external parameters of the binocular vision system to establish a conversion relation between a binocular system pixel coordinate system and a world coordinate system;

[0011] (3) solve a transformation matrix to establish a coordinate mapping relation between the SD-OCT system and the binocular vision system;

[0012] (4) use the SD-OCT system to perform orthogonal scanning on a glass surface containing a modification feature to obtain two groups of B-Scan images to construct plane equations of the upper and lower surfaces of the glass, and simultaneously use the binocular vision to shoot three-dimensional coordinates of end points of the modification feature;

[0013] (5) convert the three-dimensional coordinates of the end points of the modification feature in the binocular vision system to the SD-OCT system by the transformation matrix in step (3), solve the distances from the end points of the modification feature to the upper and lower surfaces of the glass in the SD-OCT system, and realize positioning of the glass internal modification.

[0014] In step (1), the SD-OCT system comprises a super radiation light emitting diode, a fiber coupler, a sample arm, a reference arm and a spectrometer;

[0015] The super radiation light emitting diode emits a broadband light beam which is divided into two beams after passing through the fiber coupler, one of which enters the reference arm and the other of which enters the sample arm; the light beam in the reference arm passes through a collimator and a focusing lens and then returns by a mirror; the light beam in the sample arm first passes through a collimator and then passes through a two-dimensional scanning galvanometer combined with a focusing lens to accurately scan the target glass;

[0016] The light returned from the sample arm and the reference arm is coupled again in the fiber coupler, and the coupled light beam enters the spectrometer.

[0017] In step (2), the pixel accuracy in the two directions of B-Scan and the pixel accuracy in the depth direction of A-Scan in the two-dimensional tomographic scan image of the SD-OCT system are calibrated. The specific process is as follows:

[0018] Place the checkerboard calibration board stably and accurately in the center of the field of view, ensuring that it is free from vibration or displacement throughout the operation. After scanning the calibration board with the SD-OCT system to obtain the B-Scan image of the calibration board, locate the feature points in the image, and then compare the location results with the actual physical size of the calibration board to obtain the pixel accuracy of the B-Scan image in the X and Y directions.

[0019] A reflector is placed on the sample arm of the SD-OCT system, meaning that both the sample arm and the reference arm are now reference arms. A high-precision electric displacement stage is used to move the reflector at equal distances to obtain the depth direction pixel accuracy of the A-Scan image.

[0020] In step (2), the intrinsic parameters of the binocular vision system include camera focal length, principal point coordinates, and distortion coefficients, while the extrinsic parameters include the rotation matrix and translation vector between the two cameras.

[0021] The specific process of step (3) is as follows:

[0022] The SD-OCT system and the binocular vision system each capture feature points within the same area. Let the first... The coordinates of each feature point in the SD-OCT system and the binocular vision system are respectively and The coordinates of the two systems satisfy the coordinate mapping: ,in, Let be the transformation matrix. , ~ There are 12 parameters to be estimated;

[0023] calculate ~ When mapping coordinates Decomposed into three-dimensional spatial components: ;

[0024] in, , , , Construct a matrix for the homogeneous coordinates of the calibrated feature points in a binocular vision system;

[0025] The SD-OCT system and the binocular vision system acquire feature points at four or more locations at the same position, and solve for the parameters. ~ This establishes the coordinate mapping relationship between the SD-OCT system and the binocular vision system.

[0026] In step (4), based on the depth information contained in the two groups of B-Scan images obtained from the upper and lower surfaces respectively, the plane equations of the upper and lower surfaces of the glass are constructed by the eigenvalue decomposition method based on the covariance matrix, so as to describe the spatial positions of the upper and lower surfaces of the glass in the SD-OCT system coordinate system.

[0027] The eigenvalue decomposition method based on the covariance matrix is used to construct the plane equations of the upper and lower surfaces of the glass, and the specific process is as follows:

[0028] The two-dimensional galvanometer scanning of the SD-OCT system obtains orthogonal B-Scan images, and the B-Scan scanning paths along the X-axis and the Y-axis should pass through the midpoints of each other; the straight line equation of one surface of the glass is extracted When scanning in one direction, the scanning images on the other side are all selected as the pixel positions at the center of the image, so as to obtain three-dimensional coordinates ; wherein, ) respectively represent the coordinates of the current scanning dimension in the scanning process of the SD-OCT system, and ) represents the center point coordinates of the other orthogonal dimension in the scanning process of the SD-OCT system.

[0029] The B-Scan image coordinates of two directions are obtained, a three-dimensional point cloud data set is constructed, and the eigenvalue decomposition method based on the covariance matrix is used to obtain the plane equation of the glass surface ;

[0030] The plane equations of the upper and lower surfaces of the glass are obtained by the above method.

[0031] The specific process of step (5) is as follows:

[0032] The three-dimensional point mapping of the projection of the modified feature bottom end in the binocular system is multiplied by the transformation matrix R to be converted to the SD-OCT coordinate system, so as to obtain the three-dimensional coordinates of the projection in the SD-OCT coordinate system ), and the distance from the end point of the modified feature in the glass to the lower surface of the glass is solved ;

[0033] Similarly, the three-dimensional point mapping of the projection of the modified feature top end in the binocular system is multiplied by the transformation matrix R to be converted to the SD-OCT coordinate system, so as to obtain the three-dimensional coordinates of the projection in the SD-OCT coordinate system, and finally the distance from the end point of the modified feature in the glass to the upper surface of the glass is solved, so as to position the modified feature.

[0034] Compared with the prior art, the present application has the following beneficial effects:

[0035] 1. Existing single detection technologies have significant shortcomings: On the one hand, a standalone SD-OCT system struggles to simultaneously measure surface and internal modified features of the glass (when incident perpendicularly, the backscattered light from the modified features is much weaker than the reflected light from the glass surface, making simultaneous identification impossible; when incident at an angle, it cannot image the upper and lower surfaces of the glass); on the other hand, while a standalone binocular vision system can capture images of modified features, it struggles to accurately obtain the spatial positions of the upper and lower surfaces of the glass, and cannot calculate the depth coordinates of the modified features. This invention solves the problems of "surface position measurement" and "modified feature positioning" simultaneously through a collaborative scheme of "SD-OCT constructing the plane equations of the upper and lower glass surfaces + binocular vision locating the endpoints of the modified features + coordinate transformation matrix unifying the data from both systems," achieving simultaneous and accurate measurement of both.

[0036] 2. This invention can be adapted to industrial online testing scenarios, eliminating the need for offline sampling and significantly improving testing efficiency and coverage. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram of the SD-OCT system and the binocular vision system in an embodiment of the present invention.

[0039] Figure 2 This is a schematic diagram of a scan using the SD-OCT system in an embodiment of the present invention.

[0040] Figure 3 This is a diagram showing the effect of scanning the calibration board when using the SD-OCT system and the binocular vision system in a joint calibration process according to an embodiment of the present invention.

[0041] Figure 4 This is a schematic diagram of the detection process for detecting the modified morphology inside the glass in an embodiment of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] It should be noted that, unless otherwise specified, the features in the following embodiments and implementation methods can be combined with each other.

[0044] A glass internal modification positioning method based on SD-OCT and binocular vision, comprising the following steps:

[0045] Step S1: Ensure the spatial alignment of both the frequency domain optical coherence tomography (SD-OCT) system and the binocular vision system. As shown in the figure, the basic system of the detection device is composed of a superluminescent diode 1, a fiber coupler 2, a sample arm, a reference arm, a spectrometer 7, and binocular vision systems 13 and 14. Figure 1

[0046] In this embodiment, the superluminescent diode 1 of the SD-OCT system emits a broadband light beam with a bandwidth of 140 nm, which is divided into two beams after passing through the fiber coupler 2, one entering the reference arm and the other entering the sample arm. In the reference arm, the light beam is collimated and shaped by the polarization controller 3 and the collimator 4, then focused and dispersion compensated by the focusing lens 5, and then returned by the mirror 6. In the sample arm, the light beam also passes through the polarization controller 8 and the collimator 9 first, then is accurately scanned on the sample 12 by a two-dimensional scanning galvanometer 10 (X-direction scanning and Y-direction scanning) combined with a focusing lens 11. The light returned from the sample arm and the reference arm is coupled again in the fiber coupler 2, forming a low-coherence interference effect. The coupled light beam enters the spectrometer 7, which includes a collimator, a grating, and a lens. The grating separates different wavelengths of light on the spectrum, and the lens focuses the light beam after the grating on the different positions of the line array camera according to different wavelengths. The detector converts the received light intensity signal into an electrical signal and transmits it to the connected computer. The computer then processes the signal through fast Fourier transform (FFT) to complete the conversion from the spectral domain signal to the frequency domain signal. Thus, the depth profile of the sample (A-Scan) is reconstructed. Combining multiple A-Scan data sets, a complete two-dimensional tomographic image (B-Scan) can be generated, showing the lateral and longitudinal structure of the sample.

[0047] The binocular vision system 13 is built on the left side of the two-dimensional scanning galvanometer 10 and the focusing lens 11 in the reference arm of the SD-OCT system, and the binocular vision system 14 is built on the right side. The angles between the optical axes of the two lenses of the binocular vision system and the horizontal plane are set to 45° and 75°, respectively, to satisfy the sufficient parallax angle and baseline distance of the binocular vision system 13 and the binocular vision system 14 to obtain two images of the same target. The center of the field of view and the scanning center of the SD-OCT system are coincident, and they are in the coplanar state in three-dimensional directions, so that they have the same detection range and detect the sample 12 together.

[0048] Step S2: Convert the pixel and relative coordinates of the system into actual physical scales.

[0049] ​Step 2.1: Calibrate the pixel accuracy of the two directions of the B-Scan image of the SD-OCT system.

[0050] In this example, a black and white chessboard calibration plate with a size of 1 mm is selected as the standard reference for SD-OCT calibration. The SD-OCT system is used to scan the calibration plate, and multiple sets of A-Scan images and complete B-Scan images are obtained. The pixel values corresponding to the features in the two directions of the B-Scan image are determined, as shown in Figure 2 , which are the X and Y directions. The pixel accuracy in the X-Y direction is obtained by dividing 1 mm by the number of features in the two directions. In this example, the pixel accuracy in the X and Y directions is 2.08 μm and 3.12 μm, respectively.

[0051] Step 2.2: Calibrate the pixel accuracy of the depth direction of the A-Scan image of the SD-OCT system.

[0052] For the longitudinal direction of the SD-OCT system, as shown in Figure 2 , it is the pixel calibration in the Z direction. In this example, a lifting experiment is used for verification. The sample stage in the sample arm is moved to change the optical path difference between the reference arm and the sample arm. A mirror is installed on the sample stage of the sample arm, i.e., both the sample arm and the reference arm are mirrors. The sample stage is lifted by 10 μm each time, the longitudinal spectral line information of each column is read, and the corresponding peak intensity coordinates are obtained. Each group is moved 10 times, the linearity is tested, and the pixel accuracy in the Z direction is calibrated. In this example, the physical size of one pixel in the depth direction in air is 2.755 μm.

[0053] Step 2.3: Calibrate the internal and external parameters of the binocular system, and establish the conversion relationship between the pixel coordinate system and the world coordinate system.

[0054] The right camera of the binocular system is rotated at different angles to the calibration plate, and 16-24 pictures at different angles and poses are collected for calibration. Based on the matching points in the two groups of binocular cameras, the initial values of the relative external parameters of the binocular system are solved, and the internal parameter matrix, distortion coefficient, and binocular external parameter of the left and right cameras are jointly optimized to obtain the three-dimensional data calibration result in binocular vision.

[0055] Step S3: Establish the coordinate mapping relationship of the two systems to realize the coordinate unification of the two systems.

[0056] Step 3.1: Use the SD-OCT system to scan the glass calibration plate to obtain the B-Scan image, and screen the obtained B-Scan image to obtain the position of the positioning feature point.

[0057] In this step, the SD-OCT system is controlled to scan a pre-set circular glass calibration plate. The calibration plate is a standard circle with a center-to-center distance of 300 μm, made entirely of glass (refractive index 1.525), and has a thickness of 500 μm, which is on the same order of magnitude as the glass layer thickness of the modified feature to be measured. The acquired images are then quality-screened, retaining only valid images with complete features and clear imaging, such as... Figure 3 The images shown demonstrate the imaging effects of the binocular camera and the SD-OCT system on the calibration plate. A shape-matching feature extraction algorithm identifies preset positioning feature points from the edge contours. Based on the lateral pixel accuracy of the B-Scan image calibrated in step two, the two-dimensional coordinates (X1, Y1) of the aforementioned positioning feature points in the SD-OCT system coordinate system are calculated, and the depth coordinates Z1 in the SD-OCT system are obtained by combining the scanned A-scan image.

[0058] Step 3.2: Use the binocular vision module to synchronously photograph the same area of ​​the glass calibration plate. Process the acquired left and right views through the coordinate transformation relationship obtained in step 2.3, identify and locate the same feature points as in step 3.1, and then calculate the three-dimensional coordinates of these feature points in the binocular vision system coordinate system.

[0059] Step 3.3: Construct a solvable system of linear equations for the same position coordinates obtained from the two systems above, and solve for the transformation matrix of the two systems.

[0060] In this example, during the coordinate unification process between the stereo system and the SD-OCT system, a homogeneous transformation matrix R is introduced to describe the spatial mapping relationship between the two systems: The first three rows and three columns of this matrix represent rotation transformations, and the fourth column represents translation vectors. The homogeneous form ensures the compatibility of joint operations of translation and rotation transformations, enabling three-dimensional coordinate transformations to be uniformly achieved through matrix multiplication.

[0061] In this embodiment, during the process of unifying the coordinate systems of the binocular vision system and the SD-OCT system, the dual systems identify the first [missing information] in steps 3.1 and 3.2. The coordinates of the feature points in the OCT system and the stereo system are denoted as follows: and According to the homogeneous transformation rule, the coordinates of the two systems satisfy: .

[0062] Expand and separate the translation-related terms, decomposing the coordinate mapping into three-dimensional spatial components: ;

[0063] In the formula, , , , H is a homographic matrix of the calibrated feature points in binocular vision system: ;

[0064] The transformation matrix needs to satisfy the rotation orthogonality and the translation linear superposition, which contains 12 parameters to be estimated. Each calibrated feature point can provide 3 independent equations. According to the solvability theory of linear equations, when the number of equations is not less than the number of parameters to be estimated, the system has a unique solution (or least squares optimal solution). Therefore, at least 4 non-coplanar points are needed to make the construction matrix full rank, ensuring the unique solvability of the transformation matrix parameters. Therefore, at least 4 sets of feature points for the same position need to be collected by the SD-OCT system and the binocular vision system during the calibration process.

[0065] Step S4: As shown in Figure 4 , the SD-OCT system 15 is used to perform orthogonal scanning on the upper and lower glass surfaces of the glass modification feature to obtain two B-Scan images, so as to construct the plane equation of the upper and lower surfaces 17 of the glass.

[0066] In this example, the linear equation of the upper and lower surfaces of the glass is solved, and the two orthogonal B-Scan images are obtained by two-dimensional scanning of the SD-OCT system. The Canny operator is used to extract the edges of the upper and lower surfaces of the image, and the least squares method is used to fit to obtain the values of the slope k and the intercept b of the upper and lower surfaces. The coordinates are obtained by scanning the X direction and the Y direction images respectively. In one direction scanning, the image on the other side is selected as the pixel position at the center of the image, that is, the three-dimensional coordinates , are obtained. For the obtained coplanar point cloud data, the centroid is calculated first, wherein , , The point cloud is centralized, that is, the coordinates of the centroid are subtracted to obtain a new point cloud . Based on the centralized point cloud, a 3x3 covariance matrix is calculated to describe the distribution dispersion and direction correlation of the point cloud in three-dimensional space:

[0067] ;

[0068] The covariance matrix is decomposed, the characteristic equation is solved, the eigenvalue and the corresponding eigenvector are obtained. Since the point cloud is coplanar, the minimum eigenvalue , and the corresponding eigenvector is the normal vector of the surface plane of the glass calibration plate. Combined with the centroid and the normal vector, the glass surface equation can be arranged as .

[0069] At the same time, the control binocular vision system 13, binocular vision system 14 to the modified feature 16 end point inside the glass synchronous shooting, collection get contains the left and right view of the modified feature; based on the pixel coordinate system and world coordinate system conversion relationship constructed by the calibrated binocular vision system internal participation external parameter in step 2.3, through the binocular stereo matching method, the corresponding pixel position of the modified feature end point in the left and right view is calculated by matching the left and right view, and then the three-dimensional coordinates of the end point in the binocular vision system coordinate system are solved.

[0070] Step five: through the transformation matrix in step S3, the three-dimensional coordinates of the modified feature end point in the binocular vision system are converted into the SD-OCT system combined with the linear transformation matrix R solved in step S3, the three-dimensional point coordinates of the binocular vision system are converted into the coordinate system of the SD-OCT, and the corresponding coordinates . The distance from the point to the plane is solved in the same SD-OCT coordinate system, that is, the distance from the modified feature end point to the lower surface of the glass is solved .

[0071] Similarly, the distance from the modified feature end point inside the glass to the upper surface of the glass can be solved, so as to locate the position of the modified feature.

[0072] The above embodiments have described the technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the present application. Any modification, supplement and equivalent replacement made within the principle range of the present application shall be included in the protection scope of the present application.

Claims

1. A glass interior modification positioning method based on SD-OCT and binocular vision, characterized in that, The method comprises the following steps: (1) calibrating the spatial alignment of the SD-OCT system and the binocular vision system, so that the field of view of the two systems is overlapped on the target glass region; (2) calibrating the pixel accuracy of the two directions of the B-Scan image of the SD-OCT system and the pixel accuracy of the depth direction of the A-Scan image; at the same time, calibrating the intrinsic parameters and extrinsic parameters of the binocular vision system, and establishing the conversion relationship between the pixel coordinate system and the world coordinate system of the binocular system; (3) solving the transformation matrix to establish the coordinate mapping relationship between the SD-OCT system and the binocular vision system; (4) using the SD-OCT system to obtain two groups of B-Scan images by orthogonal scanning of the glass surface containing the modified feature, so as to construct the plane equations of the upper and lower surfaces of the glass; the binocular vision simultaneously shoots the three-dimensional coordinates of the end points of the modified feature; (5) converting the three-dimensional coordinates of the end points of the modified feature in the binocular vision system to the SD-OCT system through the transformation matrix in step (3), and solving the distance from the end points of the modified feature to the upper and lower surfaces of the glass in the SD-OCT system, so as to realize the positioning of the internal modification of the glass.

2. The method of claim 1, wherein the method is a method of interior modification positioning of glass based on SD-OCT and binocular vision. In step (1), the SD-OCT system comprises a super radiation light emitting diode, a fiber coupler, a sample arm, a reference arm and a spectrometer; The super radiation light emitting diode emits a broadband light beam which is divided into two beams after passing through the fiber coupler, one of which enters the reference arm and the other of which enters the sample arm; the light beam in the reference arm passes through a collimator and a focusing lens, and then returns by a mirror; The light beam in the sample arm first passes through a collimator, and then passes through a two-dimensional scanning galvanometer combined with a focusing lens to accurately scan the target glass; The light returned from the sample arm and the reference arm is coupled again in the fiber coupler, and the coupled light beam enters the spectrometer.

3. The glass internal modification and positioning method based on SD-OCT and binocular vision according to claim 2, characterized in that, In step (2), the pixel accuracy of the two directions of the B-Scan image of the SD-OCT system and the pixel accuracy of the depth direction of the A-Scan image are calibrated, and the specific process is as follows: Place the chessboard calibration plate steadily and accurately at the center of the field of view, and ensure that it has no vibration or displacement during the entire operation process; use the SD-OCT system to scan the calibration plate to obtain the B-Scan image of the calibration plate, then locate the feature points in the image, and then compare the positioning results with the actual physical size of the calibration plate, so as to obtain the pixel accuracy of the B-Scan image in the X direction and the Y direction; Place a mirror in the sample arm of the SD-OCT system, that is, the sample arm and the reference arm are both reference arms at this time, and use a high-precision electric displacement table to move the mirror at equal distances to obtain the depth direction pixel accuracy of the A-Scan image.

4. The method of claim 1, wherein the method is a method of interior modification positioning of glass based on SD-OCT and binocular vision. In step (2), the intrinsic parameters of the binocular vision system include camera focal length, principal point coordinates and distortion coefficients, and the extrinsic parameters include rotation matrix and translation vector between the two cameras.

5. The SD-OCT and binocular vision based interior glass modification positioning method of claim 1, wherein, The specific process of step (3) is as follows: The SD-OCT system and the binocular vision system respectively shoot feature points in the same area, and the coordinates of a feature point in the SD-OCT system and the binocular vision system are respectively and ; the coordinates of the two systems satisfy coordinate mapping: wherein, , is 12 to-be-estimated parameters;​​​ Computing At this time, the coordinates are mapped into three-dimensional spatial components: ;​ wherein , , , is a matrix for constructing homogeneous coordinates of a calibration feature point in a binocular vision system; The SD-OCT system and the binocular vision system collect feature points at the same positions in four groups or more, and parameters are solved , so as to establish the coordinate mapping relationship of the SD-OCT system and the binocular vision system.​ 6. The SD-OCT and binocular vision based interior glass modification positioning method of claim 1, wherein, In step (4), based on the depth information contained in the two groups of B-Scan images obtained from the upper and lower surfaces respectively, the plane equations of the upper and lower surfaces of the glass are constructed by the feature decomposition method based on the covariance matrix, which describes the spatial position of the upper and lower surfaces of the glass in the SD-OCT system coordinate system.

7. The SD-OCT and binocular vision based interior glass modification positioning method of claim 6, wherein, The plane equation of the upper surface and the lower surface of the glass is constructed based on the eigenvalue decomposition method of the covariance matrix, and the specific process is as follows: The two-dimensional galvanometer scanning of the SD-OCT system obtains orthogonal B-Scan images, and the B-Scan scanning paths along the X axis and the Y axis should pass through the midpoints of each other When scanning in one direction, the image of the other side scanning is selected as the pixel position of the image center, that is, three-dimensional coordinates are obtained ; wherein, wherein, (x, y) represents the coordinate of the current scanning dimension in the SD-OCT system scanning process, (x, y) represents the coordinate of the current scanning dimension in the SD-OCT system scanning process, (x, y) represents the coordinate of the current scanning dimension in the SD-OCT system scanning process, The B-Scan two-direction scanning image coordinates are acquired, a three-dimensional point cloud data set is constructed, and a feature decomposition method based on a covariance matrix is used to obtain a glass surface plane equation ; The plane equation of the upper surface and the lower surface of the glass is obtained by the above method.

8. The method of claim 7, wherein the method is a method of interior modification positioning of glass based on SD-OCT and binocular vision. The specific process of step (5) is as follows: The binocular system shoots the three-dimensional point mapping of the modified feature bottom end projection, multiplies the transformation matrix R to convert to the SD-OCT coordinate system, and obtains the three-dimensional coordinates of the modified feature bottom end projection in the SD-OCT coordinate system ), and the distance from the end point of the modified feature inside the glass to the lower surface of the glass is solved . Similarly, the binocular system shoots the three-dimensional point mapping of the projection of the modified feature top, multiplies the transformation matrix R to convert to the SD-OCT coordinate system, obtains the three-dimensional coordinates of the modified feature top in the SD-OCT coordinate system, and finally solves the distance from the modified feature top to the upper surface of the glass to locate the modified feature position.

Images