Camera calibration method
By using checkerboard calibration plate and corner point detection algorithm in camera calibration, combined with homography matrix and anti-distortion iteration method, the problem of camera calibration method falling into local minimum value in nonlinear optimization in the prior art is solved, and high Accurate and robust camera parameter estimation, suitable for a variety of camera types.
Patent Information
- Application Number
- CN202411021475.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-11-22
AI Technical Summary
Existing camera calibration methods are prone to fall into local minimums in nonlinear optimization, especially in strong distortions, and traditional methods require complex three-dimensional calibration scenario deployment and insufficient robustness and accuracy.
The checkerboard calibration plate is used to gradually determine the camera's internal parameters, external parameters and distortion parameters through corner point detection, unidirectional matrix estimation, minimizing reprojection error and anti-distortion iteration steps, and improve the accuracy and robustness of calibration.
Achieve high accuracy and robust camera parameter estimation, suitable for images of different angles and positions, reduce the number of nonlinear optimization parameters, avoid local minimum values, and be suitable for ordinary and special types of cameras, improving calibration accuracy and versatility.
Smart Images

Figure CN119006609A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of photogrammetry, and in particular to a camera calibration method. Background Art
[0002] The camera self-calibration method mainly estimates the camera parameters by using the constraints of camera motion or the constraints in the scene. For example, the camera parameters are estimated by calculating the vanishing point using the parallel lines or orthogonal information in the scene. The self-calibration method is flexible and can calibrate the camera online. However, since it is based on absolute quadratic curves or surfaces, its algorithm has poor robustness.
[0003] The traditional camera calibration method requires the use of a calibration object with known size, such as a chessboard or a circular calibration plate. By establishing the correspondence between the points on the calibration object with known coordinates and its image points, a certain algorithm is used to obtain the internal and external parameters of the camera model. According to the different calibration objects, they can be divided into three-dimensional calibration objects and planar calibration objects. Three-dimensional calibration objects can be calibrated by a single image, and the calibration accuracy is high, but the processing and maintenance of high-precision three-dimensional calibration objects are more difficult. Planar calibration objects are simpler to make than three-dimensional calibration objects, and the accuracy is easy to ensure, but two or more images must be used during calibration. In the traditional calibration method, because the external parameters obtained by linear solution are not accurate enough, the external parameters are usually optimized together in nonlinear optimization. Although this can be solved in one step, it will make the optimization problem non-convex, that is, there are multiple local minima in the optimization function, which may cause the optimization to fall into the local minimum. In the case of weak distortion, using the external parameters obtained by linear solution as the initial value in nonlinear optimization can avoid the optimization from falling into the local minimum, but it is not applicable in the case of strong distortion.
[0004] Zhang Zhengyou calibration method: Zhang Zhengyou calibration method is a method between traditional calibration methods and self-calibration methods. It uses a printed checkerboard calibration plate, shoots images of the calibration plate at different angles, and uses the corner information in the image to solve the internal and external parameters of the camera. Patent CN113920201A proposes a calibration method based on epipolar geometric constraints. However, this calibration method requires first establishing an indoor three-dimensional calibration field and measuring the artificial landmarks in the field, which makes the deployment operation inconvenient. Summary of the invention
[0005] In order to solve the above problems, the present invention provides a camera calibration method, which inherits the usability characteristics of Zhang Zhengyou's calibration method. It only needs to shoot a chessboard image to automatically calibrate, and is easy to operate. The chessboard calibration plate is used for camera calibration. The camera's intrinsic parameters, extrinsic parameters and distortion parameters are accurately determined through steps such as corner point detection, homography matrix estimation, minimization of reprojection error, distortion correction, and anti-distortion iteration. Specifically, the following steps are included:
[0006] S1, take multiple chessboard images with the camera to be calibrated;
[0007] S2, extracting sampling points through corner point detection, where the corner points correspond to the intersection points of the chessboard;
[0008] S3, read the sampling points, and use the homography matrix to estimate the camera parameters, which include: camera intrinsic parameters and camera extrinsic parameters;
[0009] S4, construct a method to minimize the reprojection error and estimate the camera distortion parameters;
[0010] S5, if the reprojection error obtained in S4 is small, output the camera parameters, otherwise enter S6;
[0011] S6. Perform distortion correction on the sampling points, and repeat S3-S5.
[0012] Furthermore, S2 specifically includes: using corner point detection to extract corner points from the original checkerboard image in S1 collected by the user as sampling points, and saving the extracted sampling points as a designated file for subsequent processing.
[0013] Furthermore, the camera intrinsic parameters and camera extrinsic parameters are estimated by using the homography matrix in S3 as follows:
[0014] The homography matrix H is defined as:
[0015] H=sK(r1,r2,t)
[0016] Among them, it is a 3*3 matrix, s is the scale scaling factor, K is the camera intrinsic parameter matrix, r1, r2 are the first and second columns of the extrinsic parameter rotation matrix respectively, and t is the displacement vector of the extrinsic parameter; m=HP is substituted into different sampling points for solution, and the sampling point is denoted as m(u, v), u and v are the coordinates of the normalized point, and the object point P(X, Y, 0) is set as the position of the chessboard sampling point in the world coordinate system. After solving the homography matrix, substitute the constraints of the rotation matrix to solve the camera intrinsic parameters and the camera extrinsic parameters respectively.
[0017] Furthermore, the construction of minimizing the reprojection error in S4 is specifically as follows:
[0018]
[0019] Among them, α, β are the distortion parameters in the extended unified camera model, the initial values α=0, β=1 are set, and the parameter range is limited to α∈[0,1], β≥0, m ij is the sampled point in the i-th row and j-th column, is the camera model, R j ,t j , is the distance of pixel offset, P ijis the position of the pixel on the original chessboard, n is the number of rows of the chessboard corner points, and k is the number of columns of the chessboard corner points.
[0020] Furthermore, the distortion correction of the sampling points in S6 is specifically performed as follows:
[0021] The sampling points are subjected to the anti-distortion iteration using the camera distortion parameters estimated in S4, the camera intrinsic parameters and camera extrinsic parameters are solved again, and the estimated camera distortion parameters are updated to improve the accuracy of parameter calibration.
[0022] Furthermore, the anti-distortion iteration is specifically: using the updated estimated camera distortion parameters to anti-distort the original sampling points, so that the reprojection error between the original sampling points and the reprojected object points is minimized.
[0023] Furthermore, the anti-distortion iteration specifically includes the following steps:
[0024] S61, performing anti-distortion iteration according to the camera distortion parameters obtained in S4, substituting the camera intrinsic parameters as fixed values, solving the camera extrinsic parameters to minimize the reprojection error, and updating the camera distortion parameters;
[0025] S62, substituting the camera extrinsic parameters obtained in the previous step and the updated camera distortion parameters as fixed values, solving the camera intrinsic parameters to minimize the reprojection error, and updating the camera distortion parameters;
[0026] S63, substituting the camera intrinsic parameters obtained in the previous step and the updated camera distortion parameters as fixed values, solving the camera extrinsic parameters again to minimize the reprojection error, and updating the camera distortion parameters;
[0027] S64, loop S62-S63 until the reprojection error between the original sampling point and the reprojected object point is minimized.
[0028] Furthermore, the checkerboard image captured in S1 is a checkerboard calibration plate, which includes checkerboards of known size and position.
[0029] Furthermore, the camera intrinsic parameters in S3 include: focal length, principal point coordinates, and the camera extrinsic parameters include: rotation matrix and translation vector.
[0030] The beneficial effects of the present invention are:
[0031] 1. High accuracy: By using the checkerboard calibration plate and corner detection algorithm, feature points for parameter estimation can be accurately extracted. The method of minimizing the reprojection error can accurately estimate the camera's intrinsic parameters, extrinsic parameters, and distortion parameters, thereby improving the accuracy of camera calibration.
[0032] 2. Strong robustness: This method can adapt to checkerboard images of different angles and positions, so it has strong robustness. Even in some complex scenes, such as illumination changes, noise interference, etc., this method can obtain relatively accurate camera parameter estimation results. It effectively reduces the parameters that need to be estimated in nonlinear optimization, transforms the problem into a convex optimization problem, and improves the robustness of calibration.
[0033] 3. Iterative optimization: Introducing the iterative optimization process, performing distortion correction on the sampling points and re-estimating the camera parameters can effectively reduce the impact of distortion on parameter estimation and improve the accuracy of calibration. The camera intrinsic and extrinsic parameters are gradually iterated alternately, gradually approaching the optimal value, and the camera distortion parameters are continuously updated. Through continuous iteration, the reprojection error can be gradually reduced until the requirements are met or the maximum number of iterations is reached. Using anti-distortion iteration, the camera internal and external parameters and distortion parameters are alternately optimized, which effectively reduces the possibility of parameter optimization falling into the local minimum and increases the robustness and versatility of calibration.
[0034] 4. Wide range of applications: This method is not only applicable to ordinary cameras, but also to special types of cameras such as wide-angle cameras and fisheye cameras. It can provide accurate camera parameters for various computer vision tasks such as 3D reconstruction, target positioning, augmented reality, etc.
[0035] 5. Easy to implement: This method is based on mature image processing algorithms and mathematical optimization methods, and is easy to implement and program. The implementation process can be simplified by using open source computer vision libraries. The basic framework of this method can be easily extended to other types of calibration plates or feature point extraction algorithms. The algorithm can be optimized and improved according to specific needs to adapt to different application scenarios and camera types. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 A flowchart of a camera calibration method is shown;
[0037] Figure 2 The chessboard pictures taken at different angles by the camera to be calibrated in the embodiment;
[0038] Figure 3 A checkerboard image in which sampling points are extracted by corner point detection in the embodiment;
[0039] Figure 4 Schematic diagram of the anti-distortion iterative process in the embodiment;
[0040] Figure 5 Schematic diagram of the distortion image correction result in the embodiment. DETAILED DESCRIPTION
[0041] The following is combined with Figure 1-5The preferred embodiments of the present invention are described in detail so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.
[0042] A camera calibration method comprises the following steps:
[0043] S1, take multiple chessboard images with the camera to be calibrated;
[0044] S2, extracting sampling points through corner point detection, where the corner points correspond to the intersection points of the chessboard;
[0045] S3, read the sampling points, and use the homography matrix to estimate the parameters, including: camera intrinsic parameters and camera extrinsic parameters;
[0046] S4, construct a method to minimize the reprojection error and estimate the camera distortion parameters;
[0047] S5, if the reprojection error obtained in S4 is small, output the camera parameters, otherwise enter S6;
[0048] S6. Perform distortion correction on the sampling points, and repeat S3-S5.
[0049] Specifically, first, a camera to be calibrated is used to take multiple checkerboard images from different angles and positions. The checkerboard calibration plate usually contains checkerboards of known size and position, which are used for subsequent feature point extraction and camera parameter estimation.
[0050] Then, corner points are extracted from the checkerboard image as sampling points by a corner detection algorithm (such as Harris corner detection, FAST, etc.). These corner points correspond to the intersection points of the checkerboard and are used for subsequent parameter estimation.
[0051] Then, the homography matrix is used for parameter estimation, and the bidirectional projection process of the distorted camera model and the Jacobian matrix of the model parameters for the projection point coordinates are defined. Homography matrix: describes the mapping relationship from one plane to another. In camera calibration, it describes the mapping from the checkerboard plane (world coordinate system) to the image plane (pixel coordinate system). Camera intrinsic parameters: including focal length, principal point coordinates, etc., are used to describe the internal properties of the camera. Camera extrinsic parameters: including rotation matrix and translation vector, which are used to describe the relative position and posture between the camera coordinate system and the world coordinate system. By substituting different sampling points and corresponding object points (points on the checkerboard), the homography matrix H can be solved. Using the decomposition of the homography matrix, the intrinsic and extrinsic parameters of the camera can be solved.
[0052] Next, we construct a method to minimize the reprojection error. By using the scale-normalized root mean square error of the reprojection error, we can effectively compare the accuracy of camera calibration under different data sets. Reprojection error: refers to the error between the points in the three-dimensional space projected onto the two-dimensional image plane through the camera parameters and the actual detected points. Distortion parameters: used to describe the image distortion caused by lens distortion (such as radial distortion, tangential distortion, etc.). By constructing an objective function that minimizes the reprojection error, and using the optimization algorithm to solve the distortion parameters.
[0053] Then, check the reprojection error. If the reprojection error obtained in the previous step is small (less than a preset threshold), the camera parameter estimation is considered accurate and the camera parameters are output. Otherwise, perform distortion correction.
[0054] Distortion correction: Use the distortion parameters estimated above to perform anti-distortion processing on the original sampling points to eliminate or reduce the influence of distortion on parameter estimation. Solve the internal and external parameters of the corrected sampling points again and estimate the distortion parameters, and iterate until the reprojection error meets the requirements or reaches the maximum number of iterations.
[0055] The specific steps include:
[0056] S61, performing anti-distortion iteration according to the camera distortion parameters obtained in S4, substituting the camera intrinsic parameters as fixed values, solving the camera extrinsic parameters to minimize the reprojection error, and updating the camera distortion parameters;
[0057] S62, substituting the camera extrinsic parameters obtained in the previous step and the updated camera distortion parameters as fixed values, solving the camera intrinsic parameters to minimize the reprojection error, and updating the camera distortion parameters;
[0058] S63, substituting the camera intrinsic parameters obtained in the previous step and the updated camera distortion parameters as fixed values, solving the camera extrinsic parameters again to minimize the reprojection error, and updating the camera distortion parameters;
[0059] S64, loop S62-S63 until the reprojection error between the original sampling point and the reprojected object point is minimized.
[0060] The traditional method is to solve the camera extrinsic parameters and the camera intrinsic parameters at the same time, which will cause the parameter optimization to fall into a local minimum deadlock. For example, assuming that the camera extrinsic parameters are represented by A and B, and the camera intrinsic parameters are represented by C and D, the traditional method solves ABCD at one time, and ends the iteration when one of the values of ABCD is optimal. This method causes the parameter optimization to fall into a local minimum deadlock. The present application optimizes the parameters alternately. First, A and B are substituted as fixed values to solve the optimal solution of C and D, and the distortion parameters are updated. Then, the optimal solution of C and D obtained in the previous step is substituted as the fixed value and the updated distortion parameters to solve the optimal solution of A and B, and the distortion parameters are updated again. The solution is repeated alternately until the error meets the requirements or the maximum number of iterations is reached.
[0061] The specific process of the anti-distortion iteration is to use the estimated distortion parameters to perform anti-distortion processing on the original sampling points, so as to minimize the reprojection error between the original sampling points and the reprojected object points. This process can be achieved through an iterative optimization algorithm. Each iteration will update the distortion parameters and recalculate the reprojection error until the error meets the requirements or the maximum number of iterations is reached. Using anti-distortion iteration, the camera internal and external parameters and distortion parameters are optimized alternately, which effectively reduces the possibility of parameter optimization falling into the local minimum and increases the robustness and versatility of the calibration.
[0062] Through this method, the camera's intrinsic parameters, extrinsic parameters and distortion parameters can be accurately estimated, providing accurate camera parameters for subsequent 3D reconstruction, target positioning and other tasks.
[0063] Example:
[0064] First, if Figure 2 As shown in the figure, the user uses the camera to be calibrated to take several chessboard pictures at different angles. Picture requirements: the resolution should be consistent with the subsequent application distortion, there should be no obvious shadows and reflections, and the pictures should be taken at different angles.
[0065] Next, sampling points are extracted through corner detection. The number of rows and columns of sampling points in each image is specified by the user, such as Figure 3 As shown, it is 5*5. The module will then save the extracted sampling points as a specified file.
[0066] Next, read the sampling points and perform parameter estimation:
[0067] (1) Solving internal and external parameters
[0068] This step uses the homography matrix to construct a system of linear equations to solve the internal and external parameters.
[0069] The sampling point is m(u,v), the number of columns of the corner points of each sampling map is k, the number of rows of the corner points of the sampling map is n, and the total number of sampling points is n*k. Set the object point P(X,Y,0) as the position of the sampling point of the chessboard in the world coordinate system, which corresponds one-to-one to the sampling point on each sampling map.
[0070] The homography matrix H is defined as:
[0071] H=sK(r1,r2,t) (1)
[0072] It is a 3*3 matrix, s is the scale scaling factor, K is the camera intrinsic matrix, r1, r2 are the first and second columns of the extrinsic rotation matrix, t is the displacement vector of the extrinsic parameter, and because it is scale invariant in homogeneous coordinates, it has only 8 degrees of freedom. Each sampling image corresponds to a homography matrix, which is solved by substituting m=HP into different sampling points.
[0073] After solving the homography matrix, we can substitute equation (1) into the constraints of the rotation matrix to solve the intrinsic and extrinsic parameters respectively.
[0074] (2) Distortion Parameter Estimation
[0075] In this step, the task of minimizing the reprojection error is constructed:
[0076]
[0077] The above formula is solved by the Levenberg-Marquardt method. α, β are the distortion parameters in the extended unified camera model. The initial values α=0, β=1 are set, and the parameter range is limited to α∈[0,1], β≥0, m ij is the sampled point in the i-th row and j-th column, is the camera model, R j ,t j , is the distance of pixel offset, P ij is the position of the pixel on the original chessboard, n is the number of rows of the chessboard corner points, and k is the number of columns of the chessboard corner points.
[0078] While estimating the distortion parameters, the correction of the internal reference principal points u0 and v0 is also added to effectively improve the calibration accuracy. The overall parameters that need to be estimated are α, β, u0, and v0, which further ensures the calibration accuracy.
[0079] Then, the reprojection error is checked. If the reprojection error is small (less than a preset threshold), the camera parameter estimation is considered accurate and the camera parameters are output. Otherwise, distortion correction is performed.
[0080] Considering that when there is strong radial distortion in the image, the camera internal and external parameters solved above will have large errors, which will further affect the estimation accuracy of the distortion parameters. Therefore, the present invention designs an anti-distortion iteration, after the sampling points are anti-distorted by the distortion parameters estimated in the previous step, the internal and external parameters are solved again and the distortion parameters are estimated, thereby improving the parameter calibration accuracy.
[0081] like Figure 4 As shown in the figure, the red point array is the original sampling points, the green point array is the object points after reprojection, and the blue point array is the result of de-distorting the original sampling points using the estimated distortion parameters. The goal of minimizing the reprojection error is to make the red and green points coincide as much as possible. Through the continuous alternating iteration of the estimated distortion parameters of the blue point array and the camera intrinsic parameters and camera extrinsic parameters, it can be seen in the figure that as the de-distortion iteration steps proceed, the re-projection error is rapidly decreasing, the de-distortion result is becoming more and more linear, and the de-distortion iteration has a significant improvement on the calibration accuracy. Figure 5 The figure shows the result of distortion image correction.
[0082] The implementation results are shown in the following table:
[0083] tool loss Distortion parameter Opencv::calibrateCamera 1.614104 5 Opencv::Fisheye::calibrate 0.258343 4 Proposed 0.253071 2
[0084] In the embodiment, the number of anti-distortion iterations of the present invention is 3 times.
[0085] Among them, the Loss evaluation index is the normalized root mean square error (RMSE) of the reprojection error, the physical meaning of Loss is the average pixel error of each sampling point on the unit chessboard area, and the Loss unit is one ten-thousandth of a pixel / (sampling point*chessboard).
[0086] Through this method, the camera's intrinsic parameters, extrinsic parameters and distortion parameters can be accurately estimated, providing accurate camera parameters for subsequent 3D reconstruction, target positioning and other tasks.
[0087] Any example of the present invention can be used as an independent technical solution or combined with other examples. All patents and publications mentioned in the specification of the present invention indicate that these are public technologies in the field and can be used in the present invention. All patents and publications cited here are also listed in the references, just like each publication is specifically cited separately. The present invention here can be implemented in the absence of any element or elements, one limitation or multiple limitations, and this limitation is not specifically stated here. The terms and expressions used here are descriptive methods, but not limited by them. There is no intention to indicate that these terms and explanations described in this book exclude any equivalent features, but it can be known that any appropriate changes or modifications can be made within the scope of the present invention and the claims. It can be understood that the embodiments described in the present invention are embodiments and features in some embodiments, and any person skilled in the art can make some changes and variations based on the essence of the present invention, and these changes and variations are also considered to belong to the scope of the present invention and the scope limited by the independent claims and the appended claims.
Claims
1. A camera calibration method, It is characterized in that The specific steps include: S1, take multiple chessboard images with the camera to be calibrated; S2, extracting sampling points through corner point detection, where the corner points correspond to the intersection points of the chessboard; S3, read the sampling points, and use the homography matrix to estimate the camera parameters, which include: camera intrinsic parameters and camera extrinsic parameters; S4, construct a method to minimize the reprojection error and estimate the camera distortion parameters; S5, if the reprojection error obtained in S4 is small, output the camera parameters, otherwise enter S6; S6. Perform distortion correction on the sampling points, and repeat S3-S5.
2. A camera calibration method according to claim 1, It is characterized in that Specifically, S2 includes: using corner point detection to extract corner points from the original checkerboard image in S1 collected by the user as sampling points, and saving the extracted sampling points as a designated file for subsequent processing.
3. A camera calibration method according to claim 1, It is characterized in that In S3, the camera intrinsic parameters and camera extrinsic parameters are estimated by using the homography matrix as follows: The homography matrix H is defined as: H=sK(r 1 ,r 2 ,t) Among them, it is a 3*3 matrix, s is the scale scaling factor, K is the camera intrinsic parameter matrix, r 1 ,r 2 are the first and second columns of the extrinsic rotation matrix respectively, and t is the displacement vector of the extrinsic parameter; m=HP is substituted into different sampling points for solution, and the sampling point is denoted as m(u,v), u and v are the coordinates of the normalized point, and the object point P(X,Y,0) is set as the position of the chessboard sampling point in the world coordinate system. After solving the homography matrix, the constraints of the rotation matrix are substituted into the solution to obtain the camera intrinsic parameters and the camera extrinsic parameters respectively.
4. A camera calibration method according to claim 1, It is characterized in that The construction of minimizing the reprojection error in S4 is specifically as follows: Among them, α, β are the distortion parameters in the extended unified camera model, the initial values α=0, β=1 are set, and the parameter range is limited to α∈[0,1], β≥0, m ij is the sampling point in the i-th row and j-th column, is the camera model, R j ,t j , is the distance of pixel offset, P ij is the position of the pixel on the original chessboard, n is the number of rows of the chessboard corner points, and k is the number of columns of the chessboard corner points.
5. A camera calibration method according to claim 1, It is characterized in that The distortion correction of the sampling points in S6 is specifically performed as follows: The sampling points are subjected to the camera distortion parameters estimated in S4 for dedistortion iteration, the camera intrinsic parameters and camera extrinsic parameters are solved again, the estimated camera distortion parameters are updated, and the accuracy of parameter calibration is improved.
6. A camera calibration method according to claim 5, It is characterized in that The de-distortion iteration is specifically: de-distorting the original sampling points using the updated estimated camera distortion parameters, so that the reprojection error between the original sampling points and the reprojected object points is minimized.
7. A camera calibration method according to claim 6, It is characterized in that The anti-distortion iteration specifically comprises the following steps: S61, performing anti-distortion iteration according to the camera distortion parameters obtained in S4, substituting the camera intrinsic parameters as fixed values, solving the camera extrinsic parameters to minimize the reprojection error, and updating the camera distortion parameters; S62, substituting the camera extrinsic parameters obtained in the previous step and the updated camera distortion parameters as fixed values, solving the camera intrinsic parameters to minimize the reprojection error, and updating the camera distortion parameters; S63, substituting the camera intrinsic parameters obtained in the previous step and the updated camera distortion parameters as fixed values, solving the camera extrinsic parameters again to minimize the reprojection error, and updating the camera distortion parameters; S64, loop S62-S63 until the reprojection error between the original sampling point and the reprojected object point is minimized.
8. A camera calibration method according to claim 1, It is characterized in that The checkerboard image captured in S1 is a checkerboard calibration plate, which includes checkerboards of known size and position.
9. A camera calibration method according to claim 1, It is characterized in that The camera intrinsic parameters in S3 include: focal length, principal point coordinates, and the camera extrinsic parameters include: rotation matrix and translation vector.
Citation Information
Patent Citations
Two-step optimization method for camera parameter calibration
CN108269289A
Camera calibration method, device and equipment and computer readable medium
CN114549652A
Visual repositioning method fusing live-action three-dimensional and video
CN117726687A
Sammer camera calibration method, device and system
CN117911541A
Method and system for calibrating binocular camera, and electronic device and storage medium
WO2023045147A1
Cited By
Image distortion correction method and device, electronic equipment and storage medium
CN119295358A
Infrared irradiation point position identification method, system and computer program
CN119379814A
Large support vision measurement method based on image splicing and data fusion
CN120031984A
Rapid binocular camera calibration method
CN120259443A
A fast binocular camera calibration method
CN120259443B
