A target track field experiment calibration method based on high-precision total station

Abstract

The application discloses a target track field experiment calibration method based on a high-precision total station, first, three sets of binocular camera stations are erected, the fire line is aligned and the focus is adjusted; the internal parameters of the first set of camera stations and the rotation and translation matrix of the right camera relative to the left camera are calibrated; the calibrated camera is used to shoot a static checkerboard, 10 evenly distributed and non-collinear corner points are selected, and binocular coordinates of the 10 corner points in the left camera coordinate system are calculated; the total station establishes a north-east ground coordinate system, and global coordinates of the 10 points are measured; the binocular coordinates and the global coordinates are subjected to weighted point cloud registration, and the rotation matrix and the translation vector are solved through singular value decomposition; the above steps are repeated to complete the calibration of the remaining two sets of camera stations, and the multi-station coordinate systems are unified, and are used for solving target motion parameters. The application is flexible in deployment, high in precision, and suitable for target track field experiments.

Description

Technical Field

[0001] This invention belongs to the field of camera calibration technology / computer vision, specifically involving a target path field test calibration method based on a high-precision total station. Background Technology

[0002] Orthogonal shadow photography is an effective method for measuring the motion parameters of high-speed flying targets. The system consists of a high-resolution camera, reflective screens, and a pulsed laser. Two reflective screens and two cameras are arranged on either side of the target path, with the screens orthogonally positioned and perpendicular to the camera's optical axis. However, orthogonal shadow photography systems suffer from complexity, difficulty in debugging, and inflexible deployment. A more efficient and convenient testing method is needed in practical applications.

[0003] Binocular vision measurement systems utilize two imaging devices with identical parameters to simultaneously capture images of the same object under test, obtaining two two-dimensional images from different perspectives. The three-dimensional information of the object is then obtained by calculating the pixel difference between the two images. Compared to orthogonal shadow photography systems, binocular vision measurement systems offer advantages such as simple system structure, high efficiency, and low debugging difficulty, making them widely applicable in ballistic measurement. Furthermore, in target trajectory field experiments, the experimental environment is often complex. Binocular vision measurement systems can better adapt to the experimental environment, allowing for flexible deployment and calibration. After calibration, multiple experiments can be conducted simply by maintaining the relative positional relationship. Testing high-speed flying target parameters using binocular vision involves multiple binocular camera stations. Different binocular camera stations establish different three-dimensional coordinate systems, resulting in experimental data that cannot be directly used, and data processing is complex and labor-intensive. The calibration method proposed in this invention can unify multiple camera coordinate systems, obtaining a complete three-dimensional coordinate sequence of target motion, and features high measurement accuracy, flexible deployment, and convenience. Therefore, this application proposes a target trajectory field experiment calibration method based on a high-precision total station. Summary of the Invention

[0004] The purpose of this invention is to propose a target track field test calibration method based on a high-precision total station.

[0005] The technical solution to achieve the objective of this invention is: a target track field test calibration method based on a high-precision total station, comprising the following steps:

[0006] Step 1: Set up 3 binocular camera stations, align the binocular camera field of view with the line of fire, i.e. the expected trajectory line of the ballistics, and adjust the camera focal length. Then set up a total station to ensure that objects within the field of view of the binocular cameras are clearly visible and that the total station can measure the target points within the field of view.

[0007] Step 2: Calibrate the first set of binocular camera stations by taking pictures of the checkerboard calibration in different poses and importing them into the binocular calibration toolbox in the computer software to obtain the intrinsic parameter matrix of the camera and the rotation matrix and translation vector of the right camera relative to the left camera. The rotation matrix and translation vector are the extrinsic parameter matrix.

[0008] Step 3: Take another set of still chessboard images with the calibrated stereo camera, perform corner detection and parameter calculation, select 10 corner points that are evenly distributed, fully covered and non-collinear on the chessboard plane, extract their pixel coordinates and calculate their coordinates in the left camera coordinate system, and finally obtain the stereo coordinates of the 10 corner points, which are the three-dimensional coordinates in the left camera coordinate system.

[0009] Step 4: Use the total station to establish a northeast-east coordinate system, where X is the north coordinate, Y is the east coordinate, and Z is vertically downward. Point the laser at the 10 corner points selected in Step 3 and measure the three-dimensional coordinates of these corner points in the total station coordinate system. Use the three-dimensional coordinates in the total station coordinate system as the global coordinate system. Perform weighted point cloud registration on the binocular coordinates and the global coordinates to solve for the rotation matrix and translation vector of the coordinate system transformation. If the determinant of the rotation matrix is ​​negative one, recalculate the rotation matrix using the reflection correction method and update the translation vector accordingly.

[0010] Step 5: Repeat steps 2 to 4 to complete the calibration of the other two sets of camera stations, and finally realize that the coordinate system of the three binocular camera stations is the same as the coordinate system established by the total station, and complete the calibration work of the target track field experiment.

[0011] Furthermore, three sets of binocular vision measurement systems were set up, and a total station was installed to ensure that objects within the field of view of the binocular cameras were clearly visible and that the total station could measure any point within the field of view. This included the following steps:

[0012] Step S11: Use 6 high-resolution CCD cameras of the same model that meet the test requirements. Fix two cameras on a bracket to form a binocular vision measurement system. Set up a total station outside the camera's field of view.

[0013] Step S12: Adjust the position of the binocular camera to align the binocular camera's field of view with the fire line, and adjust the focal length of each camera in turn to ensure that objects within the binocular camera's field of view are clearly visible.

[0014] Step S13: Turn on and level the total station. Establish a global coordinate system with the center of the total station as the origin. Use the east coordinate as the Y coordinate, the north coordinate as the X coordinate, and the vertical downward direction perpendicular to the ground as the Z coordinate, i.e., the northeast-northeast coordinate system. Then measure any target point in the space to verify that the total station can work normally and acquire valid data.

[0015] Furthermore, a synchronization trigger is used to control two cameras to simultaneously capture chessboard images in different poses. The acquired images are imported into a computer's dual-camera positioning toolbox, images with large errors are removed, and parameter calculations are performed. Finally, the camera's intrinsic and extrinsic parameter matrices are exported, including the following steps:

[0016] Step S21: A person holds a chessboard and stands in the center of the camera's field of view, constantly changing the chessboard's posture, including rotating, tilting, and translating. Each time the posture is changed, a set of chessboard images is taken using a binocular camera, for a total of 30 sets.

[0017] Step S22: Import the obtained images into the binocular calibration toolbox for feature recognition and stereo matching. After matching, remove some images with large errors, including images with an average pixel reprojection error greater than 0.5 pixels. Then export the binocular camera calibration data, including the intrinsic parameter matrix and the rotation matrix and translation vector of the right camera relative to the left camera.

[0018] Furthermore, two cameras simultaneously capture calibration images of the static checkerboard pattern. Corner detection is performed on the images, extracting the pixel coordinates of 10 evenly distributed corner points. Parameter calculations are then performed to determine the coordinates of the selected points in the left camera coordinate system, i.e., the binocular coordinates. This process includes the following steps:

[0019] Step S31: Arrange the checkerboard pattern at the center of the field of view of the first set of camera stations, take a picture of the checkerboard pattern, perform corner detection on the picture, detect all corner points of the checkerboard pattern, select 10 corner points that are evenly distributed on the checkerboard pattern plane, fully cover and are not collinear, and extract their pixel coordinates.

[0020] Step S32: The obtained pixel coordinates are transformed into coordinates in a coordinate system with the left camera as the origin. The basic principle is to use singular value decomposition to solve the linear overdetermined equation system to obtain the three-dimensional coordinates of the selected point in the left camera coordinate system.

[0021] Furthermore, the total station is used to obtain the coordinates of the selected points in the total station coordinate system as global coordinates. By performing point cloud registration between the binocular coordinates and the global coordinates, the transformation relationship from the left camera coordinate system to the total station coordinate system is obtained and the accuracy is verified. The calibration of the target track field experiment is completed, including the following steps:

[0022] Step S41: Set the total station to measurement mode, select the standard station setting method, and establish a northeast-east coordinate system with the center of the total station as the origin, the north coordinate as the X coordinate, the east coordinate as the Y coordinate, and the vertical downward as the Z coordinate. Then level the total station and use the total station to control the crosshair center to align with the selected 10 points in sequence to obtain the three-dimensional coordinates of the points.

[0023] Step S42: Register the coordinates in the left camera coordinate system with the 3D coordinates obtained by the total station to obtain the rigid body transformation matrix for the transformation from the binocular coordinate system to the total station coordinate system. The basic principle is as follows: To align two sets of weighted corresponding points in the least squares sense, first differentiate the objective function with respect to the translation vector and set the derivative to zero. The translation vector is equal to the weighted center of the target point set minus the rotation matrix multiplied by the weighted center of the source point set. Substituting this translation expression back into the objective function, and expanding it while ignoring the constant term, the minimization problem is equivalent to maximizing the sum of weighted point-to-point products. This is then transformed into maximizing the trace of the matrix and constructing the weighted covariance matrix S. Then, singular value decomposition is performed on the S matrix to obtain the left singular vector matrix U and the right singular vector matrix V. The optimal rotation matrix is ​​R = VU. T Finally, the rotation matrix and translation vector give the optimal transformation relationship between the two coordinate systems;

[0024] Step S43: Substitute the transformation matrix into the binocular coordinate system and calculate its coordinates in the global coordinate system. Compare and analyze its accuracy with the three-dimensional coordinates obtained by the total station. If there is a large deviation, repeat the above steps until the absolute error σ ≤ 0.5 mm.

[0025] Furthermore, step 5 specifically includes repeating steps 2 to 4 to complete the calibration of the other two sets of camera stations, obtaining the transformation relationship, thereby realizing the transformation of multiple camera coordinate systems to a global coordinate system, and completing the target field field experiment calibration work, including the following steps:

[0026] Step S51: For the second group of binocular camera stations, repeat steps 1 to 4 of claim 1 to obtain the rotation matrix and translation vector of the second group of camera stations to the total station coordinate system.

[0027] Step S52: For the third group of binocular camera stations, repeat steps 1 to 4 of claim 1 to obtain the rotation matrix and translation vector of the third group of camera stations to the total station coordinate system.

[0028] Step S53: Transform the pixel coordinates of the three camera stations to the total station coordinate system using their respective transformation relationships to unify the coordinate systems of the three binocular camera stations.

[0029] Compared with existing technologies, the significant advantages of this invention are: 1) Quick and convenient measurement. The binocular vision measurement system only requires setting up the instrument in a suitable position and leveling it before measurement can begin quickly. This calibration method can unify the coordinate systems of each camera station, obtain the complete coordinate sequence of the target's motion process, and complete the calibration in just 50 minutes, significantly reducing the debugging time compared to orthogonal shadow photography systems, achieving "immediate measurement upon setup"; 2) High accuracy. Binocular vision has an inherent advantage in depth information analysis. In addition, the least squares method used in this invention for point cloud registration can greatly improve the accuracy of depth estimation; 3) Suitable for field experiments. The target track field experiment environment is relatively complex. Orthogonal shadow photography systems require strictly orthogonal optical paths and the measurement field of view is usually small, requiring multiple camera stations to relay the measurement. However, multiple binocular vision measurement systems combined with this calibration method can easily handle large field of view measurements, and the installation and debugging are relatively simple. Attached Figure Description

[0030] Figure 1 This is a system flowchart of the present invention.

[0031] Figure 2 This is a physical model diagram of the system of the present invention.

[0032] Figure 3 This is a system composition diagram of the present invention.

[0033] Figure 4 This is a calibration diagram of the chessboard grid for the present invention.

[0034] Figure 5 This is a diagram showing the detection results of dual-target fixed-angle points according to the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation processes described herein are merely illustrative and not intended to limit the scope of this application.

[0036] like Figure 1 As shown, this invention provides a target track field test calibration method based on a high-precision total station, the steps of which are as follows:

[0037] like Figure 1 As shown, this invention provides a target track field test calibration method based on a high-precision total station, the steps of which are as follows:

[0038] Step 1: Using a cone-shaped rod as the experimental carrier, simulating a high-speed projectile in an actual field experiment, a measurement system consisting of 6 high-resolution CCD cameras is first constructed. Two cameras form a binocular stereo vision measurement system. Three sets of binocular camera stations are arranged according to the test requirements, ensuring that the field of view of the binocular cameras is aligned with the line of fire (i.e., the expected trajectory line of the projectile). Then, a total station is set up and leveled, keeping the positions of the three sets of binocular camera stations and the total station unchanged. The total station is not moved during the calibration process.

[0039] Step 2: Calibrate the first set of binocular camera stations using a 20mm×20mm checkerboard calibration board. Connect two high-resolution CCD cameras to the industrial control computer and a synchronization trigger. Control the two cameras to simultaneously photograph the checkerboard calibration board using the synchronization trigger signal. Move the calibration board within the field of view to change its orientation and acquire 25 checkerboard images with different orientations (resolution: 9344×7000). Import these images into the computer's binocular camera calibration toolbox to obtain the intrinsic and extrinsic parameter matrices of the two cameras.

[0040] Step 3: Using a 20mm checkerboard calibration board again, photograph the checkerboard calibration board with the calibrated high-resolution camera. Perform corner detection on the acquired images, calculate and generate the pixel coordinates of 88 corner points of the checkerboard. Select 10 corner points that are evenly distributed on the checkerboard calibration board. The selected corner points must be evenly distributed in space on the two-dimensional plane of the calibration board, fully cover the area, and avoid collinearity. This is to best reflect the overall position, orientation, and scale of the checkerboard. The selected corner points are shown below. Figure 5 As shown, the pixel coordinates of the corner points are extracted, and then the pixel coordinates are converted into coordinates in a coordinate system with the optical center of the left camera as the origin (stereo coordinates). The specific operation is as follows:

[0041] For any point P(X,Y,Z) in space, assume that the corresponding imaging points in the left and right cameras are P1, P2, and P3, respectively. l (u1,v1) and P r For (u2, v2), according to the principles of monocular imaging and coordinate transformation, the following relationship holds:

[0042] (1)

[0043] In the formula: Z1 and Z2 are scale factors, i.e., depth values ​​in the left / right camera coordinate system; dx1 and dy1 are the physical dimensions of each pixel of the left camera in the x and y directions, respectively; cx1 and cx2 are the pixel coordinates of the principal point of the left camera; f1 and f2 are the focal lengths of the left and right cameras, respectively; R l and R r These represent the rotation matrices for the left and right cameras, respectively. l and t rLet A represent the translation vectors of the left and right cameras. l and A r It is the camera's intrinsic parameter matrix, A l and A r The following relationship must be satisfied:

[0044] (2)

[0045] During calibration, the camera coordinate system is based on the optical center of the left camera as its origin. The rotation matrix and translation vector of the left camera are then the identity matrix and zero vector, respectively. Let R be the rotation matrix for transforming the right camera coordinate system to the left camera coordinate system. r Let R1 be the translation vector t. r Defined as t, therefore equation (1) can be simplified to:

[0046] (3)

[0047] A l and A r It can be obtained through calibration that the product of the intrinsic parameter matrix and the extrinsic parameter matrix is ​​the projection matrix M, M l M is the left projection matrix. r The right projection matrix has the following relationship:

[0048] (4)

[0049] In the formula: Representing M l The element in the i-th row and j-th column, Representing M r The element in the i-th row and j-th column, when transformed according to the above expression, has:

[0050] (5)

[0051] Expand and organize [X,Y,Z] T By substituting and eliminating variables, we obtain the following equation:

[0052] (6)

[0053] At this point, there are 4 equations and 3 unknowns. The coordinates of the points in the left camera coordinate system can be obtained by solving the overdetermined equations. Next, the singular value decomposition (SVD) method can be used to solve the equations to obtain the 3D coordinates of the 10 points. The principle is as follows:

[0054] The above equation can be simplified as follows:

[0055]

[0056] ,

[0057] The original equation is transformed into: From singular value decomposition, we can obtain:

[0058] (7)

[0059] Substituting into the equation, we get:

[0060] (8)

[0061] The three-dimensional coordinates of the point obtained by transforming the above equation are:

[0062] (9)

[0063] After the above transformation, the three-dimensional coordinates of the 10 selected points in the left camera coordinate system can be obtained.

[0064] Step 4: Set up the total station near the object to be measured. After powering on, select the azimuth-based station setup method. Use the total station as the origin, north as the X-coordinate, east as the Y-coordinate, and the Z-axis perpendicular to the XY horizontal plane, with downwards as positive. Turn on the laser pointer, select gyroscope tilt compensation, and adjust the total station's horizontal gyroscope. Use the total station to manipulate the crosshair center to align with the selected 10 points sequentially, obtaining the three-dimensional coordinates (global coordinates) of the points in the total station coordinate system.

[0065] Step 5: Register the coordinates in the camera coordinate system with the 3D coordinates obtained from the total station to obtain the rigid body transformation matrix for the transformation from the binocular coordinate system to the total station coordinate system. The basic principle is as follows:

[0066] set up and Let R and t be two corresponding sets of points in space. To find an optimal transformation matrix that optimally aligns these two sets of points in the least squares sense, we need to find a rotation matrix R and a translation vector t such that:

[0067] (10)

[0068] In the formula, w i This is the weight for each point pair. The above function is the classic objective function of the point cloud registration algorithm. Its basic principle is to find an optimal transformation matrix (including the rotation matrix R and the translation vector t) that minimizes the difference between the transformed stereo coordinates and the coordinates in the total station coordinate system. The calculation is divided into two steps:

[0069] 1. Calculate the translation vector

[0070] (1) Definition The problem is transformed into finding the minimum value of the function F(t);

[0071] (2) By taking the derivative with respect to both sides and setting the derivative to 0, we can obtain: The results are as follows:

[0072] (11)

[0073] (3) The translation vector t is obtained as:

[0074] (12)

[0075] make , Then the translation vector can be reduced to:

[0076] 2. Calculate the rotation matrix

[0077] (1) The translation vector obtained above: , bring in ,get:

[0078] (13)

[0079] (2) Let , The rotation matrix can be transformed into:

[0080] (14)

[0081] (3) Expanding the rotation matrix R and removing factors independent of R, R can be rewritten as:

[0082] (15)

[0083] (4) Convert the solution of the rotation matrix into the calculation of the trace:

[0084] (16)

[0085] in , ,

[0086] (5) Let S = XWY T Singular value decomposition (SVD) of matrix S yields: S = U∑V T The following relationship exists:

[0087] (17)

[0088] Q=V T RU, the trace of the matrix is ​​maximized if and only if Q is an identity matrix, yielding the rotation matrix. After the above calculations, the transformation relationship between the two coordinate systems can be obtained, including the rotation matrix R and the translation vector t. Using the rotation and translation matrices, the three-dimensional coordinates of the points obtained from the binocular coordinate transformation can be further obtained.

[0089] Step 6: Substitute the transformation matrix into the verification and analyze its accuracy. The principle is as follows:

[0090] (18)

[0091] Where X w The coordinates are in the total station coordinate system. Given the coordinates of the point in the left camera coordinate system, verify the accuracy of the transformation matrix using this formula. If there is a large deviation, repeat the above steps until the error σ between the two sets of coordinates is less than 0.5mm.

[0092] Step 7: After completing the calibration of a set of binocular measurement systems, replace the checkerboard calibration plate with iron rods with conical sides, take two images with a binocular camera, extract the feature points on both sides of the iron rods, use the transformation relationship to convert the two-dimensional pixel coordinates into three-dimensional coordinates, and compare them with the three-dimensional spatial coordinates measured by the total station to optimize the transformation matrix.

[0093] Step 8: Following the above steps, calibrate the other two sets of binocular camera stations. After obtaining the conversion relationship between the three binocular vision measurement systems and the total station, the three sets of measurement data can be unified to the total station coordinate system in the formal experiment to obtain the three-dimensional coordinate data of the target in space.

[0094] Six high-resolution CCD cameras were numbered 35, 36, 37, 38, 39, and 40. Cameras 35 and 37 were the two cameras in the first group of the binocular measurement system, 36 and 38 were the two cameras in the second group, and 39 and 40 were the two cameras in the third group. The three groups of cameras were positioned and fixed in place, and measurements were taken with an unfolded measuring tape to verify this calibration method. Based on the captured images, the three-dimensional coordinates in space at 110mm, 310mm, 410mm, 510mm, and 560mm were calculated. The measurement results are shown in Table 1.

[0095] Table 1. Measurement results data for three sets of camera stations

[0096]

[0097] In the experiment, a measuring tape was used as the object to be measured, with "110mm, 410mm..." representing the corresponding graduations. The actual length of the graduations 110mm and 410mm is 300mm. Using this method, the actual measured length was 299.287mm, with an error of only 0.713mm. According to the experimental data, different camera stations can effectively unify points within the field of view into the global coordinate system, with errors all within 1mm, demonstrating high accuracy. During the experiment, the extraction of pixel coordinates, the minute displacement of the camera station, and the calculation of singular value decomposition all affect the measurement accuracy.

[0098] Step 9: Record the obtained camera intrinsic parameter matrix, extrinsic parameter matrix, and the conversion relationship between each camera station and the total station to complete the target track field test calibration.

[0099] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0100] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A target track field test calibration method based on a high-precision total station, characterized in that, Includes the following steps: Step 1: Set up 3 binocular camera stations, align the binocular camera field of view with the line of fire, i.e. the expected trajectory line of the ballistics, and adjust the camera focal length. Then set up a total station to ensure that objects within the field of view of the binocular cameras are clearly visible and that the total station can measure the target points within the field of view. Step 2: Calibrate the first set of binocular camera stations by taking pictures of the checkerboard calibration in different poses and importing them into the binocular calibration toolbox in the computer software to obtain the intrinsic parameter matrix of the camera and the rotation matrix and translation vector of the right camera relative to the left camera. The rotation matrix and translation vector are the extrinsic parameter matrix. Step 3: Take another set of still chessboard images with the calibrated stereo camera, perform corner detection and parameter calculation, select 10 corner points that are evenly distributed, fully covered and non-collinear on the chessboard plane, extract their pixel coordinates and calculate their coordinates in the left camera coordinate system, and finally obtain the stereo coordinates of the 10 corner points, which are the three-dimensional coordinates in the left camera coordinate system. Step 4: Use the total station to establish a northeast-east coordinate system, where X is the north coordinate, Y is the east coordinate, and Z is vertically downward. Point the laser at the 10 corner points selected in Step 3 and measure the three-dimensional coordinates of these corner points in the total station coordinate system. Use the three-dimensional coordinates in the total station coordinate system as the global coordinate system. Perform weighted point cloud registration on the binocular coordinates and the global coordinates to solve for the rotation matrix and translation vector of the coordinate system transformation. If the determinant of the rotation matrix is ​​negative one, recalculate the rotation matrix using the reflection correction method and update the translation vector accordingly. Step 5: Repeat steps 2 to 4 to complete the calibration of the other two sets of camera stations, and finally realize that the coordinate system of the three binocular camera stations is the same as the coordinate system established by the total station, and complete the calibration work of the target track field experiment.

2. The target track field test calibration method based on a high-precision total station according to claim 1, characterized in that, Three sets of binocular vision measurement systems were set up, and a total station was installed to ensure that objects within the field of view of the binocular cameras were clearly visible and that the total station could measure any point within the field of view. This included the following steps: Step S11: Use 6 high-resolution CCD cameras of the same model that meet the test requirements. Fix two cameras on a bracket to form a binocular vision measurement system. Set up a total station outside the camera's field of view. Step S12: Adjust the position of the binocular camera to align the binocular camera's field of view with the fire line, and adjust the focal length of each camera in turn to ensure that objects within the binocular camera's field of view are clearly visible. Step S13: Turn on and level the total station. Establish a global coordinate system with the center of the total station as the origin. Use the east coordinate as the Y coordinate, the north coordinate as the X coordinate, and the vertical downward direction perpendicular to the ground as the Z coordinate, i.e., the northeast-northeast coordinate system. Then measure any target point in the space to verify that the total station can work normally and acquire valid data.

3. The target track field test calibration method based on a high-precision total station according to claim 1, characterized in that, Two cameras are used to simultaneously capture images of a checkerboard pattern in different poses using a synchronization trigger. The acquired images are then imported into a computer's dual-camera calibration toolbox. Images with significant errors are removed, and parameter calculations are performed. Finally, the camera's intrinsic and extrinsic parameter matrices are exported. The process includes the following steps: Step S21: A person holds a chessboard and stands in the center of the camera's field of view, constantly changing the chessboard's posture, including rotating, tilting, and translating. Each time the posture is changed, a set of chessboard images is taken using a binocular camera, for a total of 30 sets. Step S22: Import the obtained images into the binocular calibration toolbox for feature recognition and stereo matching. After matching, remove some images with large errors, including images with an average pixel reprojection error greater than 0.5 pixels. Then export the binocular camera calibration data, including the intrinsic parameter matrix and the rotation matrix and translation vector of the right camera relative to the left camera.

4. The target track field test calibration method based on a high-precision total station according to claim 1, characterized in that, Two cameras simultaneously capture calibration images of a static checkerboard pattern. Corner detection is performed on the images, and the pixel coordinates of 10 evenly distributed corner points are extracted. Parameter calculations are then performed to determine the coordinates of the selected points in the left camera coordinate system, i.e., the stereo coordinates. This process includes the following steps: Step S31: Arrange the checkerboard pattern at the center of the field of view of the first set of camera stations, take a picture of the checkerboard pattern, perform corner detection on the picture, detect all corner points of the checkerboard pattern, select 10 corner points that are evenly distributed on the checkerboard pattern plane, fully cover and are not collinear, and extract their pixel coordinates. Step S32: The obtained pixel coordinates are transformed into coordinates in a coordinate system with the left camera as the origin. The basic principle is to use singular value decomposition to solve the linear overdetermined equation system to obtain the three-dimensional coordinates of the selected point in the left camera coordinate system.

5. The target track field test calibration method based on a high-precision total station according to claim 1, characterized in that, Using a total station, the coordinates of selected points in the total station coordinate system are obtained as global coordinates. Point cloud registration is performed between the binocular coordinates and the global coordinates to obtain the transformation relationship from the left camera coordinate system to the total station coordinate system and verify its accuracy. This completes the calibration of the target track field experiment, including the following steps: Step S41: Set the total station to measurement mode, select the standard station setting method, and establish a northeast-east coordinate system with the center of the total station as the origin, the north coordinate as the X coordinate, the east coordinate as the Y coordinate, and the vertical downward as the Z coordinate. Then level the total station and use the total station to control the crosshair center to align with the selected 10 points in sequence to obtain the three-dimensional coordinates of the points. Step S42: Register the coordinates in the left camera coordinate system with the 3D coordinates obtained by the total station to obtain the rigid body transformation matrix for the transformation from the binocular coordinate system to the total station coordinate system. The basic principle is as follows: To align two sets of weighted corresponding points in the least squares sense, first differentiate the objective function with respect to the translation vector and set the derivative to zero. The translation vector is equal to the weighted center of the target point set minus the rotation matrix multiplied by the weighted center of the source point set. Substituting this translation expression back into the objective function, and expanding it while ignoring the constant term, the minimization problem is equivalent to maximizing the sum of weighted point-to-point products. This is then transformed into maximizing the trace of the matrix and constructing the weighted covariance matrix S. Then, singular value decomposition is performed on the S matrix to obtain the left singular vector matrix U and the right singular vector matrix V. The optimal rotation matrix is ​​R = VU. T Finally, the rotation matrix and translation vector give the optimal transformation relationship between the two coordinate systems; Step S43: Substitute the transformation matrix into the binocular coordinate system and calculate its coordinates in the global coordinate system. Compare and analyze its accuracy with the three-dimensional coordinates obtained by the total station. If there is a large deviation, repeat the above steps until the absolute error σ ≤ 0.5 mm.

6. The target track field test calibration method based on a high-precision total station according to claim 1, characterized in that, Step 5 specifically includes repeating steps 2 to 4 to complete the calibration of the other two sets of camera stations, obtaining the transformation relationship, which enables the conversion of multiple camera coordinate systems to a global coordinate system, and completing the target field field experiment calibration work, including the following steps: Step S51: For the second group of binocular camera stations, repeat steps 1 to 4 of claim 1 to obtain the rotation matrix and translation vector of the second group of camera stations to the total station coordinate system. Step S52: For the third group of binocular camera stations, repeat steps 1 to 4 of claim 1 to obtain the rotation matrix and translation vector of the third group of camera stations to the total station coordinate system. Step S53: Transform the pixel coordinates of the three camera stations to the total station coordinate system using their respective transformation relationships to unify the coordinate systems of the three binocular camera stations.

Images