Calibration method for parameters of depth camera installed on CBCT

Abstract

The invention discloses a method for calibrating parameters of a depth camera installed on a CBCT, and the method comprises the steps: S1, placing a registration tool in a C-arm machine, and carrying out the position calibration of an identification mark and a dot matrix on the registration tool; s2, a CBCT image of the registration tool is collected through a C-arm machine, and meanwhile depth information is collected through a depth camera installed on the end face of a flat panel detector of the C-arm machine; s3, identifying an identification mark in the CBCT image, and calculating according to the position calibration in S1 to obtain the position of each point of the dot matrix in the CBCT image; s4, recognizing a dot matrix in the RGB image in the depth information, and according to the mark of each point of the dot matrix, enabling the dot matrix to be in one-to-one correspondence with the position of each point of the dot matrix in the CBCT image, so as to calculate the transformation relation between the RGB image and the CBCT image in the depth information by combining the internal reference of the camera; and S5, according to the known conversion parameters of the depth camera, the external parameters of the depth camera are obtained through calculation in combination with the step S4. The method is simple and convenient to operate, and calibration of depth camera parameters is clear and concise.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method for calibrating parameters of a depth camera installed on a CBCT. Background Art

[0002] By adding a depth camera to a 3D X-ray imaging device such as a C-arm, the depth camera image can be collected synchronously during the 3D X-ray image acquisition process to obtain complex surface information above and on both sides of the patient and the surgical area, thereby realizing navigation obstacle avoidance, surgical assistance and other functions without feeling. In order to achieve the above functions, the depth camera needs to be calibrated in advance (at the factory or during maintenance) to obtain the precise posture information of the camera movement during the scanning process for surface reconstruction of the surgical area. Summary of the invention

[0003] Purpose of the invention: In view of the above-mentioned needs, the present invention proposes a method for calibrating parameters of a depth camera installed on a CBCT.

[0004] Technical solution:

[0005] The present invention provides a method for calibrating parameters of a depth camera installed on a CBCT, which is characterized by comprising:

[0006] S1. Place the registration tooling in the C-arm machine and calibrate the position of the identification marks and dot matrix on the registration tooling;

[0007] S2, collecting the CBCT image of the registration tooling through the C-arm machine, and collecting its depth information through the depth camera installed on the C-arm machine;

[0008] S3, identifying the identification mark in the CBCT image, and calculating the position of each point in the CBCT image according to the position calibration of S1;

[0009] S4, identifying the dot matrix in the RGB image in the depth information obtained in S2, and according to the marks of each dot matrix point, making a one-to-one correspondence between the positions of each dot matrix point in the CBCT image obtained in S3, thereby calculating the transformation relationship between the RGB image in the depth information and the CBCT image in combination with the camera intrinsic parameters;

[0010] S5. According to the known conversion parameters of the depth camera, combined with S4, the conversion matrix between the depth image in the depth information and the CBCT image is calculated, that is, the external parameters of the depth camera are obtained.

[0011] Specifically, the dot matrix is ​​arranged on the outer surface of the registration tool, and the identification mark is arranged at any position of the registration tool.

[0012] More specifically, the outer surface of the registration tool is set as an arc-shaped cylinder, and the arc-shaped cylinder is concentric with the C-shaped arm and has the same curvature;

[0013] When the registration tool is placed in the C-arm machine, its axis is perpendicular to the plane where the C-arm is located.

[0014] Furthermore, the alignment tool can be designed as an inner and outer sleeve structure, the identification mark is arranged on the inner sleeve of the alignment tool, the dot matrix is ​​arranged on the outer surface of the outer sleeve, and the outer sleeve is sleeved outside the inner sleeve.

[0015] Specifically, the dot matrix is ​​composed of corner points of a chessboard, different positioning marks are added to each white square of the chessboard, and identification labels are added to each corner point of the chessboard.

[0016] More specifically, in S4, the dot matrix in the RGB image in the depth information obtained in S2 is identified, and the dot matrix is ​​matched with the position of each dot matrix point in the CBCT image obtained in S3 according to the mark of each dot matrix point, as follows:

[0017] (1) Binarize each image in the RGB image sequence to obtain a binary image;

[0018] (2) identifying the mask in the binary image obtained in step (1) as a positioning mark to define the set range of each chessboard corner point and the identification label of each chessboard corner point;

[0019] (3) According to the set range of the chessboard corner points framed in step (2), the internal gradient is calculated to obtain the edge, a cross line fitting is performed on each edge pixel, the intersection of the fitted cross lines is calculated, and the corner points of the chessboard are identified, that is, the coordinates of the corner points of the chessboard are obtained;

[0020] (4) Using the identification labels of the corner points of each chessboard obtained in step (2) and the coordinates of the corner points of the chessboard obtained in step (3), one-to-one correspondence is established between the positions of the corner points of each chessboard in the CBCT image in S3.

[0021] Specifically, in S4, the transformation relationship between the RGB image in the calculated depth information and the CBCT image is calculated using a reprojection optimization method.

[0022] More specifically, the reprojection optimization method adopts a G2O nonlinear optimization model.

[0023] Specifically, in S3, the identification mark in the CBCT image is identified by using a threshold extraction method.

[0024] Specifically, in S3, the position of each point in the lattice of the CBCT image is calculated according to the position calibration of S1, as follows:

[0025] The position of each point of the lattice in the CBCT image, that is, the transformation relationship between the lattice and the CBCT image, includes the rotation matrix R cp-ci and the translation matrix T cp-ci , then for a certain voxel in the identification mark, as follows:

[0026] P ci-sp =R cp-ci P cp-sp +T cp-ci ;

[0027] Among them, P ci-sp To identify the position of the identification mark in the CBCT image; cp-sp is the actual position of the identification mark, that is, the position of the identification mark relative to the dot matrix, which is calculated based on the position calibration of S1;

[0028] The rotation matrix R is obtained by using the SVD method. cp-ci and the translation matrix T cp-ci ,include:

[0029] S31, respectively calculating the centroid of all voxels of the identification mark in the CBCT image and the centroid of all voxels of the actual identification mark;

[0030] S33. Define matrix W = p*q T / n, where p and q represent the identification mark in the de-centroided CBCT image and the actual identification mark, respectively. The SVD singular matrix decomposition is performed to obtain W = U∑V T , thus we get the rotation matrix R cp-ci =UV T and the translation matrix T cp-ci =pR cp-ci *q, that is, obtaining the transformation relationship between the dot matrix and the CBCT image.

[0031] Beneficial effects: The present invention realizes the correspondence between the CBCT image and the depth camera through the positioning marks set on the chessboard, and then the transformation relationship between the RGB image collected by the depth camera and the CBCT image can be calculated in combination with the camera internal parameters. After that, the external parameters of the depth camera can be calculated according to the conversion parameters of the depth camera combined with the depth image, thereby completing the calibration of the depth camera parameters installed on the CBCT, and then the subsequent surface reconstruction of the surgical area can be realized. This method can realize dynamic continuous calibration, determine the motion trajectory of the depth camera and even the entire C-arm; the tooling includes a variety of positioning marks, which can be observed separately by a variety of imaging devices, and all the measurement results are associated with this method to form an organic whole. The calibration result is used for the reconstruction of the surgical area surface with a large field of view, and can be combined with the CBCT image with a smaller field of view to enrich the patient's perception of the surgical area, and cooperate with the navigation equipment to realize the required functions such as anti-collision. The present invention is simple and convenient to operate, and the calibration of the depth camera parameters is clear and concise. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are merely embodiments of the present invention, and ordinary technicians in this field can obtain other drawings based on these drawings without creative work.

[0033] Figure 1 A flow chart of a method for calibrating parameters of a depth camera installed on a CBCT according to the present invention;

[0034] Figure 2 It is a structural example diagram of the alignment tooling of the present invention;

[0035] Figure 3 An example diagram of a chessboard set up on a registration tool;

[0036] Figure 4 This is the alignment relationship diagram.

[0037] Among them, 10 is a registration tool, 11 is an inner sleeve, 12 is an outer sleeve, 13 is an identification mark, and 14 is a base. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0039] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present invention should have the common meanings understood by persons having ordinary skills in the field to which the present invention belongs.

[0040] The process of the method for calibrating the position parameters of the depth camera installed on the CBCT of the present invention is as follows: Figure 1 As shown, including:

[0041] S1, placing the registration tool in the C-arm machine, the dot matrix on the outer surface of the registration tool is concentric with the C-arm of the C-arm machine and has the same curvature, setting different marks on each point in the dot matrix, and calibrating the position of the identification mark and the dot matrix on the registration tool;

[0042] In the present invention, Figure 2 As shown, the dot matrix is ​​set on the outer surface of the registration tool 10, and the identification mark 13 can be set at any position of the registration tool 10. Specifically, the outer surface of the registration tool 10 can be set as an arc cylinder, and the arc cylinder is concentric with the C-arm and has the same curvature, and the identification mark 13 is embedded in the registration tool 10.

[0043] In a specific embodiment of the present invention, the registration tool 10 may be designed as an inner and outer sleeve structure, the identification mark 13 is arranged on the inner sleeve 11 of the registration tool 10 , and the dot matrix is ​​arranged on the outer surface of the outer sleeve 12 .

[0044] In the present invention, the identification mark 13 such as a steel bead needs to be observed by the CBCT and needs to be within the CBCT field of view. However, the CBCT field of view is small and the range for embedding the steel bead is limited. Therefore, in this embodiment, the identification mark 13 can be embedded on the inner sleeve 11.

[0045] In the present invention, in order to reduce the weight and cost of the alignment tool 10, the inner sleeve 11 can be designed as a hollow cylinder, and the material can be plastic.

[0046] In the present invention, different marks may be set for each point in the dot matrix on the outer surface of the registration tool 10 so as to be distinguishable in the image.

[0047] Specifically, the dot matrix on the outer surface of the registration tool 10 can be designed as a checkerboard structure, and the corner points of the checkerboard constitute the dot matrix, such as Figure 3 More specifically, a unique positioning mark is added to each white square of the chessboard, and an identification label is added to each corner point of the chessboard.

[0048] In this embodiment, the inner sleeve 11 of the alignment tool 10 can be designed as a cylindrical structure, preferably a cylinder, on which a plurality of asymmetrically arranged identification marks 13 are inlaid. Specifically, the identification marks 13 can be made of steel balls, the number of which can be 5; the outer sleeve 12 is sleeved on the outside of the inner sleeve 11, and the portion thereof corresponding to the inner sleeve 11 on which the identification marks 13 are inlaid can be hollowed out, so that the identification marks 13 inlaid on the inner sleeve 11 are exposed to the outside, so that the position of the alignment tool 10 can be conveniently measured after it is assembled, thereby realizing the calibration between the identification marks 13 and the chessboard, that is, the position calibration between the identification marks and the dot matrix on the alignment tool can be completed by measuring the position of the dot matrix and the identification marks on the alignment tool. Of course, the position calibration between the identification marks and the dot matrix on the alignment tool can also be completed by the design parameters and installation parameters of the alignment tool 10 itself. In the present invention, the outer sleeve 12 can be designed as a hollow fan-shaped columnar structure, and the checkerboard can be formed by directly processing on the fan-shaped cylindrical surface of the outer sleeve 12, or can be obtained by printing and then pasted on the fan-shaped cylindrical surface of the outer sleeve 12.

[0049] In the present invention, in order to ensure the accuracy of calibration, a base 14 may be provided. Figure 2 As shown, after the registration tool 10 is assembled, it is fixed on the base 14. At this time, a three-coordinate measuring machine is used to calibrate the positions of the identification mark 13 and the chessboard. After the calibration is completed, the base 14 can be removed.

[0050] In the present invention, when the alignment tool 10 is placed in the C-arm machine, the dot matrix on the outer surface of the alignment tool is concentric with the C-arm of the C-arm machine and has the same curvature. At this time, the axis of the alignment tool 10 is perpendicular to the plane where the C-arm is located.

[0051] In the present invention, when the alignment tool 10 is placed in the C-arm machine, the adjustment of its position needs to be confirmed by fluoroscopic exposure. Specifically, the C-arm is slid to the front and side positions for exposure respectively, and it is confirmed that the positioning steel beads can be seen on the fluoroscopic image. At this time, the installation of the alignment tool 10 is completed.

[0052] S2, collecting CBCT images of the registration tooling through the C-arm machine, and collecting depth information through the depth camera installed on the C-arm;

[0053] In the present invention, the depth camera is installed on the C-arm at a position facing the alignment tooling 10. In this embodiment, the depth camera can be installed on the flat-panel detector end of the C-arm or on the inner wall of the C-arm of the C-arm.

[0054] S3, identifying the identification marks in the CBCT image acquired by S2, calculating the transformation relationship between the dot matrix and the CBCT image according to the position calibration of S1, and then calculating the position of each point in the dot matrix in the CBCT image;

[0055] In the present invention, a threshold extraction method can be used to identify the identification mark in the CBCT image collected by S2. In a specific embodiment of the present invention, the Hu value of the steel ball is extremely high, and a fixed threshold Tr can be set. When the Hu value of a point in the CBCT image is greater than the fixed threshold Tr, it is considered to be a voxel on the steel ball, so that the steel ball in the CBCT image can be identified.

[0056] The position of each chessboard corner point in the CBCT image is calculated based on the position calibration of S1, specifically:

[0057] Taking the dot matrix formed by the corner points of the chessboard as an example, the transformation relationship between the dot matrix, i.e. the chessboard, and the CBCT image includes the rotation matrix R cp-ci and the translation matrix T cp-ci , then for a certain voxel in the identification mark, as follows:

[0058] P ci-sp =R cp-ci P cp-sp +T cp-ci ;

[0059] Among them, P ci-sp To identify the position of the identification mark in the CBCT image; cp-sp is the actual position of the identification mark, that is, the position of the identification mark relative to the chessboard, which is calculated based on the position calibration of S1;

[0060] The rotation matrix R can be obtained by using the SVD method cp-ci and the translation matrix T cp-ci ,include:

[0061] S31, respectively calculating the centroid of all voxels of the identification mark in the CBCT image and the centroid of all voxels of the actual identification mark, as follows:

[0062]

[0063] Wherein, m represents the number of voxels in the identification mark in the CBCT image, and n represents the number of voxels in the actual identification mark;

[0064] S32, de-centrifying the identification mark in the CBCT image and the actual identification mark according to S31, as follows:

[0065]

[0066] S33. Define matrix W = p*q T / n, and perform SVD singular matrix decomposition to obtain W=U∑V T , thus we can get the rotation matrix R cp-ci =UV T and the translation matrix Tcp-ci =pR cp-ci *q, the transformation relationship between the chessboard and the CBCT image can be obtained, and then the position of each chessboard corner point in the CBCT image can be calculated.

[0067] S4, according to the RGB image in the depth information collected by the depth camera in S2, the points in the dot matrix are identified according to S1, and the positions of the points in the dot matrix in the CBCT image obtained in S3 are matched one by one, so that the transformation relationship between the RGB image collected by the depth camera and the CBCT image can be calculated in combination with the camera internal parameters;

[0068] In the present invention, the depth information collected by the depth camera in S2 includes an RGB image sequence and a depth image sequence. By identifying a certain frame image in the RGB image sequence, the checkerboard corner points in the RGB image collected by the depth camera can be obtained, and the checkerboard corner points can be matched one by one with the positions of the points in the dot matrix of the CBCT image obtained in S3, as follows:

[0069] (1) Binarize each image in the RGB image sequence to obtain a binary image;

[0070] (2) identifying the mask in the binary image obtained in step (1) as a positioning mark to define the set range of each chessboard corner point and the identification label of each chessboard corner point;

[0071] (3) according to the set range of the chessboard corner points framed in step (2), the internal gradient is calculated to obtain the edge, a cross line fitting is performed on each edge pixel, the intersection of the fitted cross lines is calculated, and the corner points of the chessboard are identified, that is, the coordinates of the corner points of the chessboard are obtained;

[0072] (4) Using the identification labels of the corner points of each chessboard obtained in step (2) and the coordinates of the corner points of the chessboard obtained in step (3), one-to-one correspondence is established between the positions of the corner points of each chessboard in the CBCT image in S3.

[0073] In the present invention, after the positions of each checkerboard corner point in the RGB image collected by the depth camera and each checkerboard corner point in the CBCT image are one-to-one corresponding, a reprojection optimization method can be used to calculate the transformation relationship between the RGB image collected by the depth camera and the CBCT image. In a specific embodiment of the present invention, the reprojection optimization method adopts a G2O nonlinear optimization model, that is, the transformation matrix H between the positions of each checkerboard corner point in the RGB image collected by the depth camera and the positions of each checkerboard corner point in the CBCT image obtained by iteratively optimizing the depth image through a gradient descent method, that is, the transformation relationship between the RGB image collected by the depth camera and the CBCT image, so that the difference between the positions of the checkerboard corner points relative to the depth camera under the same reference and the positions of each checkerboard corner points in the CBCT image is minimized.

[0074] Specifically, the difference between the positions of the checkerboard corner points relative to the depth camera and the positions of the checkerboard corner points in the CBCT image transformed to the same reference can be expressed as E=q cor (j)-M*H*P ci-cor ; Among them, q cor (j) represents the position of the checkerboard corner points identified in the depth information relative to the depth camera; P ci-cor represents the position of each chessboard corner point in the CBCT image; M represents the internal parameter of the depth camera, which is provided by the camera itself and is known; the goal of the iterative optimization of the gradient descent method is to obtain the corresponding H when |E| is less than the set threshold; then in the present invention, after each iteration, it is determined whether |E| is less than the set threshold or whether the number of iterations has reached the set number. If any of them is satisfied, the iteration is terminated to obtain the corresponding H at this time, otherwise the iterative optimization is continued with the corresponding H as the initial value.

[0075] In the present invention, the initial value of the transformation matrix between the initial positions of the checkerboard corner points relative to the depth camera and the positions of the checkerboard corner points in the CBCT image can be set to:

[0076]

[0077] Among them, θ is the rotation angle of the depth camera assembly when it is assembled, that is, the angle between the lens plane of the depth camera and the plane where the flat-panel detector of the C-arm machine is located; x, y, z are the offset parameters in the corresponding directions when the depth camera is assembled, is the sliding angle of the C-arm of the C-arm machine.

[0078] S5. According to the known conversion parameters of the depth camera, combined with the depth image in the depth information collected by the depth camera in S2 and the transformation relationship between the RGB image collected by the depth camera and the CBCT image obtained in S4, the conversion matrix between the depth image collected by the depth camera and the CBCT image is calculated, that is, the external parameters of the depth camera, such as Figure 4 As shown;

[0079] Among them, the conversion parameter of the depth camera is the conversion matrix between the RGB image and the depth image in the depth camera, which is provided by the camera itself and is known.

[0080] S6. Repeat S4 and S5 to obtain the external parameters of the depth camera corresponding to all frame RGB images and depth images in the depth information collected by the depth camera, that is, the final registration result.

[0081] The present invention realizes the correspondence between the CBCT image and the depth camera through the positioning marks set on the chessboard, and then the transformation relationship between the RGB image collected by the depth camera and the CBCT image can be calculated in combination with the camera internal parameters. After that, the external parameters of the depth camera can be calculated according to the conversion parameters of the depth camera combined with the depth image, thereby completing the calibration of the depth camera parameters installed on the CBCT, and then the subsequent surface reconstruction of the surgical area can be realized. This method can realize dynamic continuous calibration, determine the motion trajectory of the depth camera and even the entire C-arm; the tooling includes a variety of positioning marks, which can be observed by a variety of imaging devices respectively, and all the measurement results are associated to form an organic whole through this method. The calibration result is used for the reconstruction of the surgical area surface with a large field of view, and can be combined with the CBCT image with a smaller field of view to enrich the patient's perception of the surgical area, and cooperate with the navigation equipment to realize the required functions such as anti-collision. The present invention is simple and convenient to operate, and the calibration of the depth camera parameters is clear and concise.

[0082] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0083] The embodiments of the present invention are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for calibrating parameters of a depth camera installed on a CBCT, characterized in that: include: S1. Place the registration tooling in the C-arm machine and calibrate the position of the identification marks and dot matrix on the registration tooling; S2, collecting the CBCT image of the registration tooling through the C-arm machine, and collecting its depth information through the depth camera installed on the C-arm machine; S3, identifying the identification mark in the CBCT image, and calculating the position of each point in the CBCT image according to the position calibration of S1; S4, identifying the dot matrix in the RGB image in the depth information obtained in S2, and according to the marks of each dot matrix point, making a one-to-one correspondence between the positions of each dot matrix point in the CBCT image obtained in S3, thereby calculating the transformation relationship between the RGB image in the depth information and the CBCT image in combination with the camera intrinsic parameters; S5. According to the known conversion parameters of the depth camera, combined with S4, the conversion matrix between the depth image in the depth information and the CBCT image is calculated, that is, the external parameters of the depth camera are obtained.

2. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 1, characterized in that: The dot matrix is ​​arranged on the outer surface of the registration tool, and the identification mark is arranged at any position of the registration tool.

3. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 2, characterized in that: The outer surface of the registration tool is set as an arc-shaped cylinder, and the arc-shaped cylinder is concentric with the C-shaped arm and has the same curvature; When the registration tool is placed in the C-arm machine, its axis is perpendicular to the plane where the C-arm is located.

4. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 3, characterized in that: The alignment tool can be designed as an inner and outer sleeve structure, the identification mark is arranged on the inner sleeve of the alignment tool, the dot matrix is ​​arranged on the outer surface of the outer sleeve, and the outer sleeve is sleeved outside the inner sleeve.

5. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 1, characterized in that: The dot matrix is ​​composed of corner points of a chessboard, different positioning marks are added to each white square of the chessboard, and identification labels are added to each corner point of the chessboard.

6. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 5, characterized in that: In S4, the dot matrix in the RGB image in the depth information obtained in S2 is identified, and the dot matrix is ​​matched with the position of each dot matrix point in the CBCT image obtained in S3 according to the mark of each dot matrix point, as follows: (1) Binarize each image in the RGB image sequence to obtain a binary image; (2) identifying the mask in the binary image obtained in step (1) as a positioning mark to define the set range of each chessboard corner point and the identification label of each chessboard corner point; (3) According to the set range of the chessboard corner points framed in step (2), the internal gradient is calculated to obtain the edge, a cross line fitting is performed on each edge pixel, the intersection of the fitted cross lines is calculated, and the corner points of the chessboard are identified, that is, the coordinates of the corner points of the chessboard are obtained; (4) Using the identification labels of the corner points of each chessboard obtained in step (2) and the coordinates of the corner points of the chessboard obtained in step (3), one-to-one correspondence is established between the positions of the corner points of each chessboard in the CBCT image in S3.

7. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 1, characterized in that: In S4, the transformation relationship between the RGB image in the depth information and the CBCT image is calculated using a reprojection optimization method.

8. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 7, characterized in that: The reprojection optimization method adopts the G2O nonlinear optimization model.

9. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 1, characterized in that: In S3, the identification mark in the CBCT image is identified by using a threshold extraction method.

10. The method for calibrating parameters of a depth camera installed on a CBCT according to claim 1, characterized in that: In S3, the position of each point in the lattice of the CBCT image is calculated based on the position calibration of S1, as follows: The position of each point of the lattice in the CBCT image, that is, the transformation relationship between the lattice and the CBCT image, includes the rotation matrix R cp-ci and the translation matrix T cp-ci , then for a certain voxel in the identification mark, as follows: P ci-sp =R cp-ci P cp-sp +T cp-ci ; Among them, P ci-sp To identify the position of the identification mark in the CBCT image; cp-sp is the actual position of the identification mark, that is, the position of the identification mark relative to the dot matrix, which is calculated based on the position calibration of S1; The rotation matrix R is obtained by using the SVD method. cp-ci and the translation matrix T cp-ci ,include: S31, respectively calculating the centroid of all voxels of the identification mark in the CBCT image and the centroid of all voxels of the actual identification mark; S33. Define matrix W = p*q T / n, where p and q represent the identification mark in the de-centroided CBCT image and the actual identification mark, respectively. The SVD singular matrix decomposition is performed to obtain W = U∑V T , thus we get the rotation matrix R cp-ci =UV T and the translation matrix T cp-ci =pR cp-ci *q, that is, obtaining the transformation relationship between the dot matrix and the CBCT image.

Images