Sealed viewport calibration method for monocular and binocular three-dimensional scanning systems

Abstract

The invention belongs to the technical field of underwater calibration, and particularly relates to a sealed viewport calibration method of a monocular and binocular three-dimensional scanning system. Installing a sealed viewport on a three-dimensional scanner shell, and obtaining the distance from the sealed viewport close to a camera side interface to the end face of the shell; detaching a sealing viewport mounted on the shell of the three-dimensional scanner, and meanwhile, fastening the camera to ensure that the pose of the internal camera is not changed; placing the calibration plate in the view field of the camera in the air, changing the pose of the calibration plate for multiple times, and solving the internal parameters and the distortion coefficient of the single-viewpoint model of the camera; the movable guide rail is connected with the scanner shell and the calibration plate through the clamping tool, the one-dimensional movable guide rail is perpendicular to the installation face of a sealing viewport of the three-dimensional scanner shell through positioning connection, and the one-dimensional movable guide rail is perpendicular to the calibration face of the calibration plate through positioning clamping. Namely, the calibration plate moving along the one-dimensional guide rail is parallel to the sealed viewport of the three-dimensional scanner shell. According to the invention, the sealing viewport key parameter calibration of the underwater three-dimensional scanning system can be rapidly completed.

Description

Technical Field

[0001] The present invention belongs to the technical field of underwater calibration, and particularly relates to a calibration method for a sealed viewport of a monocular and binocular three-dimensional scanning system. Background Art

[0002] With the rapid development of the marine economy, the application demand for underwater three-dimensional measurement equipment is increasing. The development of three-dimensional measurement equipment with a monocular structured light or binocular structured light or pure binocular structure centered on a camera has received wide attention. The camera of such equipment must pass through a sealed window to ensure stable operation in air and water. According to the test results, the refraction error caused by a sealed viewport with a small thickness in air can be ignored, while the refraction error caused by a sealed viewport with a small thickness in water cannot be ignored. Modeling or compensating for the refraction of multiple underwater media is a major technical problem in the development of underwater three-dimensional measurement equipment. Generally, spherical or planar sealed viewports are used at home and abroad. Although the problem of establishing a multi-layer medium refraction model can be avoided for a spherical sealed viewport, it is difficult to guarantee the actual assembly accuracy and the processing difficulty of the spherical sealed viewport is large. Therefore, most underwater three-dimensional measurement equipment preferably uses a planar sealed viewport.

[0003] For a planar sealed viewport, the underwater model of the camera can be accurately established by using the method of ray tracing based on geometric optics. However, methods such as ignoring the thickness of the sealed viewport, assuming that the plane of the sealed viewport is parallel to the camera imaging plane, and approximating with a multi-order nonlinear model can only construct a camera underwater model with limited accuracy and cannot meet the requirements of actual high-precision underwater measurement. The accurate model involves the calibration of key parameters such as the distance d from the camera optical center to the sealed viewport interface and the pose [R, t] of the sealed viewport in the camera coordinate system. Some traditional calibration methods cannot be applied to the calibration of key parameters based on a simplified model, some are only applicable to the calibration of key parameters for a long-distance sealed viewport, and some are complex and inaccurate due to the coupling calibration of multiple parameters.

[0004] After retrieval, Chinese Patent 202111030950.5 discloses an underwater multi-medium refraction imaging model that ignores the glass thickness. Based on this simplified model, the Levenberg-Marquardt method is used to approximately solve one of the sealed viewport parameters. This simplified model cannot meet the high-precision measurement requirement within 1 mm. Chinese Patent 202211069443.7 discloses a multi-medium refraction model based on the assumption that the sealed viewport plane is parallel to the camera imaging plane. The actual assembly cannot guarantee parallelism, resulting in a large error in this model. Chinese Patent 202210545603.4 also discloses a multi-medium refraction model based on the assumption that the sealed viewport plane is parallel to the camera imaging plane. Similarly, the actual assembly cannot guarantee parallelism, resulting in a large error in this model. Chinese Patent 202010453260.X discloses a method for calibrating the object-image mapping relationship under multi-medium conditions applicable to light field cameras. On the one hand, the rationality and accuracy of the third-order mapping relationship need to be further confirmed. On the other hand, it is actually very difficult to ensure the movement of the calibration board along the camera optical center direction. Chinese Patent 202111060501.5 discloses a method and device for verifying the three-dimensional positioning of underwater structures in high-speed videos. In the verification method, the calibration of the medium surface is achieved by pasting artificial landmark points on the medium surface. This method is only applicable when the medium surface is within the clear imaging range of the camera and requires the medium surface to be large enough and fixed in position, and cannot solve the problem of calibrating the sealed viewport at close range for three-dimensional scanning sensors. Indian scholars Amit Agrawal et al. constructed an accurate underwater camera model based on ray tracing in the paper "A theory of multi-layer flat refractive geometry" and iteratively obtained the results according to the refraction constraint and the coplanarity constraint. Due to the large number of system parameters, mutual coupling, and different scales, it is difficult to solve the global optimal solution for multi-layer media and it is impossible to directly solve the sealed viewport parameters in one step. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for calibrating the sealed viewport of a monocular and binocular three-dimensional scanning system, which can quickly complete the calibration of the key parameters of the sealed viewport of an underwater three-dimensional scanning system, including the distance from the camera optical center to the sealed viewport interface and the pose of the sealed viewport in the camera coordinate system.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] A method for calibrating the sealed viewport of a monocular and binocular three-dimensional scanning system,

[0008] S1. Install a sealed viewport on the outer shell of the three-dimensional scanner, and obtain the distance D0 from the interface of the sealed viewport close to the camera side to the end face of the outer shell;

[0009] S2. Remove the sealed viewport installed on the outer shell of the three-dimensional scanner, and at the same time tighten the camera to ensure that the internal camera pose remains unchanged;

[0010] S3. Place the calibration board within the camera's field of view in the air. Change the pose of the calibration board multiple times to solve for the internal parameters M and distortion coefficients of the camera's single-viewpoint model.

[0011] S4. The moving guide rail is connected to the scanner housing and the calibration board respectively through the clamping tooling. Through positioning and connection, the one-dimensional moving guide rail is perpendicular to the sealed viewport mounting surface of the three-dimensional scanner housing, and through positioning and clamping, the one-dimensional moving guide rail is perpendicular to the calibration surface of the calibration board, that is, the calibration board moving along the one-dimensional guide rail is parallel to the sealed viewport of the three-dimensional scanner housing.

[0012] S5. Move the calibration board away from the camera until the calibration board fills the entire field of view. Record the position value L1 of the one-dimensional moving guide rail at this moment. This position value is the unified zero point relative to the one-dimensional moving guide rail.

[0013] S6. The camera captures the calibration board image in the current state, and respectively obtains the corresponding coordinate values of multiple fiducial points on the calibration board in the world coordinate system and the image coordinate system. Based on the series of coordinate values, as well as the camera's internal parameters and distortion coefficients, solve for the external parameters [R', t'] of the calibration board plane corresponding to the camera. [R', t'] is the pose of the sealed viewport in the camera coordinate system.

[0014] S7. Solve for the distance D1 from the camera optical center to the calibration board plane according to the camera's internal parameters M, external parameters [R', t'], and distortion coefficients.

[0015] S8. Move the calibration board closer to the camera until the calibration board is in close contact with the three-dimensional scanner housing. Record the position value L2 of the one-dimensional moving guide rail at this moment.

[0016] S9. Calculate the distance d from the camera optical center to the sealed viewport interface according to d = D1 - (L2 - L1) - D0.

[0017] Obtain the distance D0 from the interface of the sealed viewport close to the camera side to the housing end face through theoretical calculation or actual measurement.

[0018] Based on the Zhang Zhengyou calibration principle, complete the internal parameters M and distortion coefficients of the camera's single-viewpoint model.

[0019] Use the homography matrix identification to solve for the camera's external parameters [R', t'].

[0020] If there are other sealed viewports, repeat steps S1 - S9 to complete the calibration.

[0021] The beneficial effects achieved by the present invention are:

[0022] Complete the high-precision calibration of the sealed viewport parameters of the three-dimensional scanning system in the air, and solve the problem of calibrating the key parameters of the underwater model of the three-dimensional scanner camera; the calibration accuracy of the sealed viewport parameters is high. After verification, the reprojection error of the camera underwater model based on this calibration parameter is better than 0.2 pixels, meeting the requirements of underwater high-precision measurement. Description of the Drawings

[0023] Figure 1 This is the flowchart of the present invention;

[0024] Figure 2 This is the schematic diagram of the calibration device;

[0025] Figure 3 This is the distance schematic diagram. Detailed Embodiment

[0026] The present invention will be described in detail below with reference to the drawings and specific embodiments.

[0027] A method for calibrating the sealed viewport of a monocular and binocular three-dimensional scanning system is as follows:

[0028] S1. Install a sealed viewport on the outer shell of the three-dimensional scanner, and obtain the distance D0 from the interface of the sealed viewport close to the camera side to the end face of the outer shell;

[0029] S2. Remove the sealed viewport installed on the outer shell of the three-dimensional scanner, and at the same time tighten the camera to ensure that the internal camera pose remains unchanged;

[0030] S3. Place the calibration board within the camera's field of view in the air, change the pose of the calibration board multiple times, and calculate the internal parameters M and distortion coefficients of the camera's single-viewpoint model based on the Zhang Zhengyou calibration principle;

[0031] S4. The moving guide rail is respectively connected to the outer shell of the scanner and the calibration board through a clamping tooling. Through positioning connection, the one-dimensional moving guide rail is perpendicular to the installation surface of the sealed viewport of the three-dimensional scanner outer shell, and through positioning clamping, the one-dimensional moving guide rail is perpendicular to the calibration surface of the calibration board, that is, the calibration board moving along the one-dimensional guide rail is parallel to the sealed viewport of the three-dimensional scanner outer shell;

[0032] S5. Move the calibration board away from the camera until the calibration board fills the entire field of view, and record the position value L1 of the one-dimensional moving guide rail at this moment. This position value is the unified zero point relative to the one-dimensional moving guide rail;

[0033] S6. The camera captures the image of the calibration board in the current state, respectively obtains the corresponding coordinate values of multiple fiducial points of the calibration board in the world coordinate system and the image coordinate system, and based on the series of coordinate values, the camera internal parameters, and the distortion coefficients, uses the homography matrix to identify and solve the camera external parameters [R', t'], and [R', t'] is the pose of the sealed viewport in the camera coordinate system;

[0034] S7. Solve for the distance D1 from the camera optical center to the calibration plate plane based on the internal parameters M, external parameters [R', t'], and distortion coefficients of the camera;

[0035] S8. Move the calibration plate closer to the camera until the calibration plate is in close contact with the outer shell of the 3D scanner, and record the position value L2 of the one-dimensional moving guide at this moment;

[0036] S9. Calculate the distance d from the camera optical center to the sealed viewport interface according to d = D1 - (L2 - L1) - D0;

[0037] S10. If there are other sealed viewports, the calibration can also be completed by repeating steps S1 - S9 in the same way.

[0038] The calibration device includes a calibration plate 4, a moving guide 3, and a clamping tooling 2. The moving guide 3 is connected to the outer shell 1 of the scanner through the clamping tooling 2 to ensure that the moving guide 2 is strictly perpendicular to the sealed viewport 5 and the calibration plate 4.

[0039] A calibration method for the sealed viewport of a monocular and binocular 3D scanning system includes the following steps:

[0040] S1. For the sealed viewport installed on the outer shell of the 3D scanner, obtain the distance D0 from the interface close to the camera side of the sealed viewport to the end face of the outer shell through theoretical calculation based on the 3D model and 2D drawings, or through actual measurement using a coordinate measuring machine or a laser tracker;

[0041] S2. Remove the sealed viewport installed on the outer shell of the 3D scanner, and at the same time tighten the camera to ensure that the internal camera pose remains unchanged;

[0042] S3. Place the calibration plate within the camera's field of view in the air, change the pose of the calibration plate multiple times, and complete the internal parameters M and distortion coefficients of the camera's single-viewpoint model based on the Zhang Zhengyou calibration principle. The camera's single-viewpoint model is shown as follows;

[0043]

[0044] u′ = u + u(k 1 r 2 + k 2 r 4 + k 3 r 6 ) + [2p 1 v + p 2 (r 2 + 2u 2 )]

[0045] v′ = v + v(k 1 r 2 + k 2 r 4 + k 3 r6 ) + [2p 2 u + p 1 (r 2 +2v 2 )]

[0046] where [X C Y C Z C 1] T is the homogeneous coordinate of a point in the camera coordinate system, [X W Y W Z W 1] T is the homogeneous coordinate of a point in the world coordinate system, [R, t] is the rigid body transformation matrix from the world coordinate system to the camera coordinate system, M is the camera internal parameter, and (k 1 k 2 k 3 p 1 p 2 ) are the distortion coefficients.

[0047] S4. The moving guide rail is respectively connected to the scanner housing and the calibration plate through clamping fixtures. Through positioning connection, the one-dimensional moving guide rail is perpendicular to the sealed viewport mounting surface of the three-dimensional scanner housing, and through positioning clamping, the one-dimensional moving guide rail is perpendicular to the calibration surface of the calibration plate, that is, the calibration plate is parallel to the sealed viewport of the three-dimensional scanner housing along the movement of the one-dimensional guide rail;

[0048] S5. Move the calibration plate away from the camera until the calibration plate fills the entire field of view, and record the position value L1 of the one-dimensional moving guide rail at this moment. This position value is the unified zero point relative to the one-dimensional moving guide rail;

[0049] S6. The camera captures the calibration plate image in the current state, and respectively obtains the corresponding coordinate values of multiple fiducial points of the calibration plate in the world coordinate system and the image coordinate system. Based on the series of coordinate values, the camera internal parameter, and the distortion coefficient, the external parameters [R', t'] of the camera are identified and solved by using the homography matrix. [R', t'] is the pose of the sealed viewport in the camera coordinate system;

[0050] S7. Solve the distance D1 from the camera optical center to the calibration plate plane according to the camera internal parameter M, the external parameter [R', t'], and the distortion coefficient;

[0051]

[0052] Solve the coordinates (X W ', Y W ', Z W ') of the origin of the camera coordinate system, that is, the optical center (0, 0, 0) in the calibration plate world coordinate system. Then, according to the definition of the world coordinate system, D1 = |Z W '|;

[0053] S8. Move the calibration board closer to the camera until the calibration board is in close contact with the housing of the 3D scanner, and record the position value L2 of the one-dimensional moving guide rail at this moment;

[0054] S9. Calculate the distance d from the optical center of the camera to the sealed viewport interface according to d = D1 - (L2 - L1) - D0;

[0055] S10. If there are other sealed viewports, then the calibration can also be completed by repeating steps S1 - S9 in the same way.

Claims

1. A calibration method for the sealed viewport of a monocular and binocular 3D scanning system, characterized in that: S1. Install a sealed viewport on the outer shell of the 3D scanner, and obtain the distance D0 from the interface of the sealed viewport close to the camera side to the end face of the outer shell; S2. Remove the sealed viewport installed on the outer shell of the 3D scanner, and at the same time tighten the camera to ensure that the pose of the internal camera remains unchanged; S3. Place the calibration board within the field of view of the camera in the air, change the pose of the calibration board multiple times, and complete the initial values of the internal parameters M and distortion coefficients of the single-viewpoint model of the camera; S4. The moving guide rail is connected to the outer shell of the scanner through a clamping tooling. Through positioning connection, the one-dimensional moving guide rail is perpendicular to the installation surface of the sealed viewport of the 3D scanner outer shell, and the one-dimensional moving guide rail clamps the calibration board; S5. Move the calibration board away from the camera until the calibration board fills the entire field of view, and record the position value L1 of the one-dimensional moving guide rail at this moment; S6. The camera captures the image of the calibration board in the current state, respectively obtains the corresponding coordinate values of multiple fiducial points of the calibration board in the world coordinate system and the image coordinate system, and based on the series of coordinate values, the internal parameters and initial distortion coefficients of the camera, solve the external parameters [R', t'] of the camera. [R', t'] is the pose of the sealed viewport in the camera coordinate system; S7. Solve the distance D1 from the camera optical center to the calibration board plane according to the internal parameters M, external parameters [R', t'] and initial distortion coefficients of the camera; S9. Move the calibration board closer to the camera until the calibration board is in close contact with the outer shell of the 3D scanner, and record the position value L2 of the one-dimensional moving guide rail at this moment; S10. Calculate the distance d from the camera optical center to the interface of the sealed viewport according to d = D1 - (L2 - L1) - D0.

2. The calibration method for the sealed viewport of a monocular and binocular 3D scanning system according to claim 1, characterized in that: The distance D0 from the interface of the sealed viewport close to the camera side to the end face of the outer shell is obtained through theoretical calculation or actual measurement.

3. The calibration method for the sealed viewport of a monocular and binocular 3D scanning system according to claim 1, characterized in that: Based on the Zhang Zhengyou calibration principle, the internal parameters M and initial distortion coefficients of the single-viewpoint model of the camera are completed.

4. The calibration method for the sealed viewport of a monocular and binocular 3D scanning system according to claim 1, characterized in that: The external parameters [R', t'] of the camera are identified and solved by using the homography matrix.

5. The calibration method for the sealed viewport of a monocular and binocular 3D scanning system according to claim 1, characterized in that: If there are other sealed viewports, repeat steps S1 - S9 to complete the calibration.

Images