Visual measurement system calibration method fused with laser ranging

Abstract

The invention relates to the technical field of space measurement and calibration, and discloses a laser ranging-fused vision measurement system calibration method, which comprises the following steps of: S1, establishing a system coordinate model of a vision measurement system and a laser range finder, and defining a laser ranging coordinate system, a camera coordinate system and a world coordinate system; s2, calibrating a camera, and obtaining an internal parameter matrix and an external parameter matrix of the camera; s3, acquiring image pixel coordinates of the laser points on the calibration plate; s4, converting the pixel coordinates of the laser points into three-dimensional coordinates under a camera coordinate system based on the internal reference matrix and the external reference matrix of the camera; s5, fitting a direction vector of a laser beam through the three-dimensional space distribution data of the laser points; and S6, based on the laser beam direction vector and the laser ranging value. A direction vector of a laser beam is extracted through principal component analysis, abnormal points are eliminated in combination with a random sampling consistency algorithm, a fitting result is optimized, and the calibration robustness and precision are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of spatial measurement and calibration, and particularly to a calibration method for a vision measurement system integrating laser ranging. Background Art

[0002] With the rapid development of space technology, large synthetic aperture radar (SAR) antennas play an irreplaceable role in satellite remote sensing imaging. The surface shape accuracy of the SAR antenna is directly related to the quality and performance of satellite imaging. However, during the on-orbit operation of the satellite, the surface shape of the antenna is affected by multiple factors in a complex environment, especially microgravity, thermal effects, and vibration disturbances in space, which inevitably causes the antenna to deform. This deformation will lead to a decrease in the flatness of the SAR antenna surface, and further seriously affect the accuracy of remote sensing imaging.

[0003] Currently, traditional measurement methods, such as single vision measurement or laser ranging technology, although each has certain advantages in measurement accuracy or resolution, still face many problems when meeting the real-time measurement requirements of the antenna on orbit. For example, a single vision measurement method is easily affected by ambient light interference and it is difficult to achieve high-precision target measurement over a large range, while laser ranging, although having high precision, has limitations in coverage and rapid positioning. When the two technologies are used alone, it is difficult to fully meet the harsh requirements of the aerospace field for high precision, real-time performance, and full coverage monitoring.

[0004] In addition, current measurement methods usually rely on fixed conditions in a ground laboratory, such as a stable measurement environment and an ideal target surface. In a complex space environment, these methods have poor adaptability to vibration and temperature difference changes and lack robustness in a dynamic scenario. At the same time, in the overall monitoring of large and complex structures, traditional methods are limited by sensor distribution and technical means and it is difficult to achieve comprehensive data collection and comprehensive analysis. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a calibration method for a vision measurement system integrating laser ranging, which solves the technical problems of insufficient accuracy, weak anti-interference ability, and inability to meet the requirements of high precision and real-time performance of the existing calibration methods in a complex dynamic scenario.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A calibration method for a vision measurement system integrating laser ranging, comprising the following steps:

[0007] S1. Establish a system coordinate model of the vision measurement system and the laser rangefinder, define the laser ranging coordinate system, the camera coordinate system, and the world coordinate system, and clarify the spatial conversion relationship between the coordinate systems;

[0008] S2. Calibrate the camera to obtain the internal parameter matrix and external parameter matrix of the camera;

[0009] S3. Obtain the image pixel coordinates of the laser points on the calibration board;

[0010] S4. Based on the internal parameter matrix and external parameter matrix of the camera, convert the pixel coordinates of the laser points into three-dimensional coordinates in the camera coordinate system;

[0011] S5. Fit the direction vector of the laser beam through the three-dimensional spatial distribution data of the laser points;

[0012] S6. Based on the direction vector of the laser beam and the laser ranging value, calculate the external parameters of the laser rangefinder, including the rotation matrix and translation matrix, to complete the calibration.

[0013] Preferably, in the step S1, the establishment of the system coordinate model includes:

[0014] Define the laser ranging coordinate system as a three-dimensional coordinate system with the light-emitting point of the laser rangefinder as the origin and the direction of the laser beam as one of the coordinate axes;

[0015] Define the camera coordinate system as a three-dimensional coordinate system with the optical center of the camera as the origin and the optical axis direction as one of the coordinate axes;

[0016] Define the world coordinate system as a three-dimensional coordinate system with the fixed point of the calibration board as the origin;

[0017] Establish the conversion relationship between the laser ranging coordinate system and the camera coordinate system, including the rotation matrix and translation matrix, and define the conversion relationship between the camera coordinate system and the world coordinate system.

[0018] Preferably, in the step S2, the camera calibration adopts the checkerboard calibration method. By collecting checkerboard images at different positions and angles, calculate the internal parameters and external parameters of the camera. The internal parameters include the focal length, the position of the principal point, and the pixel scale factor, and the external parameters include the rotation matrix and translation matrix of the camera relative to the world coordinate system.

[0019] Preferably, the method for obtaining the image pixel coordinates of the laser points in the step S3 includes:

[0020] Perform grayscale processing on the collected image to separate the background and the laser point area;

[0021] Use the threshold segmentation method to extract the laser points;

[0022] Eliminate noise through morphological processing to obtain the accurate image pixel coordinates of the laser points.

[0023] Preferably, the method for converting the pixel coordinates of the laser points into three-dimensional coordinates in the camera coordinate system in the step S4 includes:

[0024] Using the camera intrinsic matrix, project the pixel coordinates of the laser points into the camera coordinate system;

[0025] Using the known spatial position information of the calibration board, map the laser points from the camera coordinate system to the world coordinate system through the camera extrinsic matrix;

[0026] Based on the coordinate transformation relationship of the laser rangefinder, convert the laser points from the world coordinate system back to the camera coordinate system.

[0027] Preferably, in step S5, the fitting of the laser beam direction vector adopts the principal component analysis method, which specifically includes:

[0028] Calculate the center point of the three-dimensional distribution of the laser points as the mean point;

[0029] Calculate the covariance matrix of the laser point distribution;

[0030] Perform eigenvalue decomposition on the covariance matrix, and extract the eigenvector corresponding to the largest eigenvalue as the direction vector of the laser beam.

[0031] Preferably, in the process of fitting the laser beam direction vector in step S5, the random sample consensus algorithm is used to optimize the laser point data, including:

[0032] Randomly select two points from the laser point dataset to fit a straight line;

[0033] Calculate the distance from other points to this straight line, and take the points less than the set threshold as inliers;

[0034] Repeat sampling and select the model with the largest number of inliers as the fitting result;

[0035] Use the inlier set of the best model to refit and determine the final laser beam direction vector.

[0036] Preferably, in step S6, the method for calculating the extrinsic parameters of the laser rangefinder based on the laser beam direction vector and the laser ranging value includes:

[0037] According to the laser beam direction vector, calculate the rotation matrix of the laser rangefinder relative to the camera;

[0038] Combined with the laser ranging value, calculate the translation matrix of the laser rangefinder relative to the camera.

[0039] Preferably, after step S6, the calibration result is verified by an error evaluation method, including:

[0040] Calculate the angle between the fitted direction vector of the laser beam and the true direction vector, and evaluate the angular error;

[0041] Calculate the Euclidean distance between the three-dimensional coordinates of the fitted laser points and the true coordinates, and evaluate the position error;

[0042] Through multiple data collection experiments, the error model is established and the calibration accuracy is analyzed.

[0043] A visual measurement system calibration device integrating laser ranging, comprising:

[0044] Laser rangefinder, used to generate a laser beam and measure the distance value of a target point;

[0045] A camera for acquiring an image including the laser point;

[0046] Calibration board, used to provide spatial reference information for camera and laser rangefinder;

[0047] A data processing module is connected to the laser rangefinder and the camera for processing image data and extracting the pixel coordinates of the laser point;

[0048] The operation module is connected to the data processing module and is used to calculate the external parameters of the laser rangefinder based on the laser point pixel coordinates, the laser distance measurement value and the coordinate model.

[0049] The present invention provides a method for calibrating a visual measurement system integrating laser ranging, which has the following beneficial effects:

[0050] 1. The present invention extracts the direction vector of the laser beam through principal component analysis, removes outliers by combining the random sampling consistency algorithm, optimizes the fitting results, and significantly improves the robustness and accuracy of the calibration. Compared with the method of simply calculating the direction of the laser beam in the prior art, which is difficult to resist noise interference, the present invention overcomes the influence of outliers on the fitting results and is more stable and reliable.

[0051] 2. The present invention generates multiple laser spots by moving the checkerboard plane multiple times during the calibration process, and uses the internal and external parameters of the camera calibration to accurately calculate the three-dimensional coordinates of each laser point in the camera coordinate system, and then solves the spatial direction of the laser beam through the fitting method. Compared with the traditional calibration method that can only handle a single laser point or incomplete position information, the method of the present invention makes the calibration result more accurate and can adapt to complex calibration scene requirements.

[0052] 3. The present invention uses the laser distance measurement value and the fitted laser direction vector to solve the rotation and translation matrix between the laser rangefinder and the camera with a combined mathematical model, thereby achieving high-precision registration between the two coordinate systems. The prior art usually relies on manual calibration of specific scenes or complex equipment support, and the calibration accuracy is easily limited by the environment. The present invention automatically completes the calibration through mathematical models and optimization algorithms, is simple to operate, does not rely on high-precision external equipment, and has stronger applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a coordinate schematic diagram of the present invention;

[0054] Figure 2 Schematic diagram for calibration of the present invention;

[0055] Figure 3 Schematic diagram showing the variation of the calibration error of the present invention with the image plane error;

[0056] Figure 4 Schematic diagram showing the variation of the calibration error of the present invention with the number of laser points;

[0057] Figure 5 Schematic diagram showing the variation of the calibration error of the present invention with the laser ranging accuracy;

[0058] Figure 6 Physical diagram of the fusion measurement system of the present invention;

[0059] Figure 7 Schematic diagram of the camera calibration result of the present invention. Detailed implementation manners

[0060] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0061] Please refer to the attached Figure 1 , an embodiment of the present invention provides a calibration method for a vision measurement system integrating laser ranging, including:

[0062] Establishing a system model: First, establish a mathematical model of the system integrating laser ranging and vision measurement, and define the conversion relationships between the laser ranging coordinate system, the camera coordinate system, and the world coordinate system. By establishing these relationships, theoretical support is provided for the calibration process.

[0063] As Figure 1 shown, first establish a camera coordinate system O C with the optical center of the camera as the origin, establish a pixel coordinate system O uv on the image plane, (u0, v0) being the principal point coordinates, and establish a world coordinate system O W with the corner point at the upper left corner of the checkerboard as the origin. Then establish a laser coordinate system O L with the laser emission point of the laser rangefinder as the origin, where the laser emission direction is set as O LZ , and it is agreed that O LX is perpendicular to the direction of O LZ and parallel to the O C -XZ plane, and O LYThe direction is determined by the right - hand rule. The integrated measurement system of the laser rangefinder and the camera needs to calibrate the spatial relationship between the two, that is, the transformation relationship from the laser coordinate system O L to the camera coordinate system O C [R L2C ,T L2C .

[0064] Suppose the coordinate representation of the spatial point P in the camera coordinate system is P C =(X c ,Y c ,Z c ), T , and the coordinate representation in the world coordinate system is P W =(X W ,Y W ,Z W ), T , and the homogeneous coordinate representation in the image coordinate system is p=(u,v,1) T , and the coordinate representation in the laser coordinate system is P L =(X L ,Y L ,Z L ). T . Then the transformation relationship between P C and P W can be expressed by Equation (1). Since the laser point is on the checkerboard plane, Z W is zero.

[0065]

[0066] Where [R W2C T W2C is the rotation matrix and translation matrix for converting from the world coordinate system to the camera coordinate system.

[0067] Expanding from Equation (1) gives

[0068]

[0069] The pin - hole model is used to describe the transformation relationship of the spatial point P in the camera coordinate system O C projected onto the image coordinate system p=(u,v,1) T as shown in Equation (3):

[0070]

[0071] Where the matrix K is the internal parameter matrix of the camera, f u and f νis the pixel focal length of the camera, s is the tilt coefficient, and (u0, v0) is the position of the principal point of the image. By combining equations (1) and (3), the imaging model of the camera is as shown in equation (4), where ρ is the normalization coefficient.

[0072]

[0073] According to (3) and (4), we can obtain:

[0074]

[0075] Through the image processing and camera calibration process, in equation (5), p = (u, v, 1) T and P W = (X W , Y W , Z W ) T are known quantities. Also, from the conversion relationship between the spatial point coordinates P L in the laser coordinate system and the coordinates P C in the camera coordinate system, it is:

[0076]

[0077] where is the direction vector of the laser in the camera coordinate system, d is the laser ranging value, is the coordinate of the origin of the laser coordinate system in the camera coordinate system. According to equation (6), the transformation matrix [R L2C , T L2C between the two coordinate systems can be calculated.

[0078] Specifically, this model defines the spatial conversion relationship between the laser rangefinder, the camera, and the world coordinate system, and provides the necessary theoretical support for the calibration process. Specifically, first, a camera coordinate system is established with the optical center of the camera as the origin. At the same time, a pixel coordinate system is defined on the image plane, where the principal point coordinates are used to determine the position of the optical center of the camera. Second, a world coordinate system is established with the corner point at the upper left corner of the checkerboard calibration board as the origin, which is used to provide the spatial reference information of the calibration board relative to the camera and the laser rangefinder. Finally, a laser coordinate system is established with the light-emitting point of the laser rangefinder as the origin, where the emission direction of the laser is defined as the main axis of the laser coordinate system, and the axis perpendicular to the main axis and parallel to the laser emission plane is determined by the right-hand rule.

[0079] The fusion measurement system of the camera and the laser rangefinder needs to accurately calibrate the spatial conversion relationship between the two, that is, to determine the rotation and translation matrices from the laser coordinate system to the camera coordinate system.

[0080] Mathematically, let the coordinates of a certain spatial point in the camera coordinate system be P c , and the coordinates in the world coordinate system be Pw , the homogeneous coordinate in the image coordinate system is p, and the coordinate in the laser coordinate system is P l The transformation relationship between the camera coordinate system and the world coordinate system can be expressed by the rotation matrix R cw and the translation matrix t c w represents that its transformation relationship is:

[0081] P c =R cw P w +t cw .

[0082] Since the laser point is usually located on a checkerboard plane, the components of the normal vector of the plane can be set to zero. In an ideal case, the pinhole imaging model is used to describe the transformation relationship from the spatial point in the camera coordinate system to the image coordinate system, which can be expressed as:

[0083] p=KP c ,

[0084] Among them, K is the camera's intrinsic parameter matrix, including the pixel focal length (f x and f y ), tilt coefficient (γ), and principal point position (c x ,c y ). By combining the coordinate system transformation relationship and the pinhole imaging model, the imaging model of the camera can be further derived as follows:

[0085] p=K(R cw P w +t cw ).

[0086] Through image processing and camera calibration, the intrinsic parameter K in the imaging model and the spatial position (P_w) of the checkerboard calibration point can be made known. On the other hand, the conversion relationship between the coordinate system of the laser rangefinder and the camera coordinate system is described as

[0087] P c =P l0 +dv l ,

[0088] Among them, P l0 is the position of the origin of the laser rangefinder coordinate system in the camera coordinate system, v l is the direction vector of the laser in the camera coordinate system, and d is the distance measurement value of the laser. Through the above formula, the rotation and translation matrix between the laser rangefinder and the camera coordinate system can be solved, thereby realizing the joint calibration of the laser ranging and visual test system. These mathematical models and transformation relationships provide a solid theoretical basis for subsequent laser point extraction, direction vector fitting and error analysis.

[0089] Calibration method based on laser spot straight line fitting: An embodiment of the present invention proposes a calibration method based on laser spot straight line fitting. This method uses the conversion relationship between the spatial coordinates and image coordinates of laser points, and uses the principal component analysis (PCA) algorithm to perform straight line fitting on the laser spot data to obtain the direction vector of the laser beam.

[0090] According to Figure 1 the definition of the laser ranging coordinate system in LZ the direction of the O axis is the direction vector of the required laser straight line O LX the direction of the axis satisfies the following relationship:

[0091]

[0092] From the orthonormality of the coordinate system, the direction of the O LY axis can be obtained as ,

[0093]

[0094] According to the transformation relationship between coordinate systems, the rotation matrix R L2C is:

[0095]

[0096] Therefore, it is necessary to solve the coordinates P C =(X c , Y c , Z c ) T of the laser points in the camera coordinate system, and solve by multi-point fitting straight line. Substituting into equations (6) and (9) can calibrate the rotation matrix R L2C between the two coordinate systems, and combining with the laser ranging value d can calibrate the translation matrix transformation matrix T L2C .

[0097] To obtain the image point coordinates of multiple laser points, as Figure 2 shown, during the calibration process, the checkerboard plane is moved multiple times to pose different poses, and it is ensured that a laser spot is formed on the checkerboard plane for each pose.

[0098] Using the captured checkerboard image and the Zhang Zhengyou checkerboard calibration method to calibrate the camera alone can obtain the internal and external parameters of the camera. The point image coordinates p on the checkerboard calibration plate plane and the world coordinates P W satisfy the mapping relationship:

[0099]

[0100] Among them, a0, a1, a2, b0, b1, b2, c1, c2 are mapping parameters to be solved. For each checkerboard pose view, using the corresponding coordinates of at least 4 points not on the same straight line, the 8 mapping parameters can be solved by the least squares method. After obtaining the values of the mapping parameters, substitute the pixel coordinates of the laser point into the mapping model to calculate the coordinates (X W , Y W , 0) T of the laser point P in the checkerboard world coordinate system. Then, according to Equation (1) and using the external parameters [R W2C T W2C of the camera calibration, the coordinates P C = (X c , Y c , Z c ) T of the laser point P in the camera coordinate system can be obtained.

[0101] The three-dimensional coordinates of each laser spot in the camera coordinate system are accurately calculated by the above method Then, the principal component analysis (PCA) is used for the straight-line fitting of the laser beam. The specific steps are as follows:

[0102] (1): Calculate the mean point of the laser spots:

[0103]

[0104] (2): Center and normalize all laser spots:

[0105]

[0106] (3): Calculate the covariance matrix and perform singular value decomposition:

[0107]

[0108] Perform singular value decomposition on the covariance matrix. The eigenvector corresponding to the largest eigenvalue is the direction vector of the laser beam

[0109] (4): The straight-line equation where the light beam is located is:

[0110]

[0111] (5): To improve the robustness of line fitting and avoid the influence of outliers, the Random Sample Consensus (RANSAC) algorithm can be used to optimize the laser spot data. First, randomly select two points from the laser point coordinate dataset to fit a line, then calculate the distances from all data points to this line, and consider the points with distances less than the set threshold as inliers. Repeat the above steps multiple times and select the model with the largest number of inliers as the best model. After obtaining the best model, use the inlier set to perform PCA fitting again to obtain a more accurate laser direction vector. Then substitute the laser ranging value d into Equation (6) to calculate the transformation matrix [R L2C ,T L2C to complete the calibration.

[0112] Specifically, the method steps are as follows:

[0113] 1. Definition of the laser ranging coordinate system and solution of the direction vector: According to the definition of the laser ranging coordinate system, the direction of the line formed by the laser spot is the main axis direction of the laser rangefinder, and this direction satisfies the following constraint relationships

[0114] v l ·v x =0, v l ·v y =0

[0115] where v x , v y , v z are the three orthogonal unit vectors of the laser rangefinder coordinate system, and v l =v z . After obtaining the laser beam direction vector by principal component analysis fitting, the rotation matrix R lc can be further solved according to the orthonormality of the coordinate system.

[0116] 2. Image acquisition and preparation of calibration data:

[0117] Laser point acquisition: As Figure 2 shown, during the calibration process, move the calibration plate with the checkerboard grid multiple times to pose it in different positions to ensure that multiple laser spots are formed on the calibration plate plane by the laser beam.

[0118] Camera calibration: Use the Zhang Zhengyou checkerboard calibration method to calibrate the camera to obtain the internal parameters (such as focal length, principal point position) and external parameters (rotation and translation matrices of the camera relative to the world coordinate system) of the camera.

[0119] Solution of the mapping relationship: After camera calibration, map the pixel coordinates of the checkerboard plane points to their true coordinates in the world coordinate system, and use 4 non-collinear checkerboard points to calculate the mapping parameters by the least squares method.

[0120] 3. Solution of the three-dimensional coordinates of the laser points:

[0121] Camera coordinate system solution: According to Equation (1):

[0122] World coordinate system solution: Substitute the pixel coordinates of the laser points into the mapping model to calculate their coordinates in the world coordinate system of the checkerboard calibration board.

[0123] P c = R cw P w + t cw

[0124] Convert the position of the laser points in the world coordinate system to the three-dimensional coordinates in the camera coordinate system through the external camera parameters.

[0125] 4. Linear fitting of the laser beam direction vector:

[0126] Calculation of the mean point: Calculate the mean point of the three-dimensional coordinates of the laser spot:

[0127]

[0128] Centering and normalization: Center the laser spot data and calculate the centered laser points:

[0129]

[0130] Calculation and decomposition of the covariance matrix: Calculate the covariance matrix of the laser spot data

[0131]

[0132] Perform singular value decomposition on the covariance matrix, and extract the eigenvector corresponding to the largest eigenvalue as the laser beam direction vector.

[0133] 5. Fitting of the line where the beam is located: The laser beam direction vector v l The expression of the line where the beam is located can be obtained through fitting:

[0134] L(t) = P0 + tv l

[0135] where P0 is the mean point of the laser spot, and v l is the laser beam direction vector.

[0136] 6. Optimization of the fitting result (RANSAC algorithm):

[0137] To improve the robustness of the fitting and avoid interference from abnormal points, the Random Sample Consensus algorithm (RANSAC) is used to optimize the fitting result.

[0138] In the laser spot coordinate dataset, randomly select two points to fit a straight line, calculate the distances from other points to the line, and regard the points with distances less than the set threshold as inliers.

[0139] Repeat the sampling multiple times. Select the model with the largest number of inliers in the selected area as the best model, and re - perform PCA fitting with the inlier data to finally obtain a more accurate laser direction vector.

[0140] 7. Solving the rotation and translation matrices:

[0141] Using the laser beam direction vector v l , combined with the ranging value d of the laser point, substitute into Equation (6):

[0142] P c = P l0 + dv l

[0143] Solve for the rotation matrix R lc and the translation matrix t lc between the laser rangefinder and the camera to complete the calibration between the laser rangefinder and the camera.

[0144] Experimental verification:

[0145] During the simulation process, the internal parameter matrix K of the camera is generated according to the real camera parameters. Assume that the true values of the laser direction vector and the coordinates of the light - emitting point are known. Define the checkerboard plane as a 9×9 square grid, with the side length of each square being 23 mm. In the range from 200 mm to 1200 mm away from the measurement system, randomly generate 20 checkerboard planes with different poses at equal intervals. The pitch angle of each checkerboard pose is randomly generated within the range of [-20°, 20°], and the translation amount is randomly generated within the range of [-20 mm, 20 mm]. Calculate the intersection points of the laser rays and the checkerboard plane as the laser spot coordinates Use the PCA method for straight - line fitting to obtain the direction vector of the laser beam Combined with the laser ranging value d, the external parameter calibration result of the laser rangefinder can be obtained according to Equation (6).

[0146] The experiment uses the angular error and the distance error as evaluation criteria. The angular error is the angle between the solved direction vector and the true direction vector, and the distance error is the distance deviation between the solved laser spot and the true intersection point. Compare the calibrated direction vector and the coordinates of the light - emitting point with the given true values. The results show that the angular error is 0°, and the distance error between the laser spot and the true value is 0 mm, proving that the proposed external parameter calibration method is feasible in principle.

[0147] To verify the influence of image plane error on calibration accuracy, Gaussian noise with a mean of 0 and a standard deviation ranging from 0.25 to 3.0 pixels was added to the two-dimensional image coordinates of the laser points. 10 independent experiments were conducted for each noise level, as Figure 3 shown. The experimental results show that there is approximately a linear relationship between the calibration parameters and the noise. As the image plane error increases, both the angular error and the distance error of the direction vector increase. When the standard deviation of the image plane error is 0.5 pixels, the spatial distance error of the laser point coordinates can meet the requirements of the measurement system at 0.5 mm.

[0148] To verify the influence of the number of laser points on calibration accuracy, Gaussian noise with a mean of 0 and a standard deviation of 0.25 pixels was added to simulate a real experiment, and then checkerboards with 4 to 20 different poses were generated respectively. 20 independent experiments were conducted for each pose number to obtain the calibration errors under different numbers of laser points. As Figure 4 shown, the experimental results show that the number of laser points has a relatively obvious influence on the calibration results of the direction vector. At least 10 laser points are required to obtain a relatively stable calibration result of the laser direction vector.

[0149] To verify the influence of laser ranging accuracy on calibration accuracy, Gaussian noise with a mean of 0 and a standard deviation ranging from 0.25 to 3.5 mm was added to the laser ranging values, and a phase plane error with a standard deviation of 0.25 pixels was added to simulate the real situation. 10 independent experiments were conducted for each ranging accuracy to obtain the calibration errors under different ranging accuracies. The results are as Figure 5 shown. The experimental results show that the laser ranging accuracy has a linear influence on the distance error of the calibration results and has little influence on the angular error. In addition, there are more outliers, indicating that the distance error is more sensitive to laser ranging noise but has little influence on the calibration of the laser direction vector. The measurement system's accuracy requirements can be met when the laser ranging accuracy is within 0.5 mm.

[0150] Calibration actual measurement experiment and analysis:

[0151] As Figure 6 shown, a measurement system consisting of an AVT camera and a laser rangefinder was built. During the experiment, 30 checkerboard images and the corresponding laser ranging values were collected within the range of 0.5 m to 2 m from the measurement system.

[0152] First, the camera was calibrated using the checkerboard calibration method. 29 effective checkerboard images were selected for camera calibration in the experiment. As Figure 7 shown, the average reprojection error of the calibration results has a mean value of 0.06 pixels. Thus, the internal parameter matrix K of the camera and the external parameter matrix [R W2C T W2C. Subsequently, these images are corrected for distortion using the calibration results to prepare for subsequent joint calibration.

[0153] In the joint calibration process, image processing is first performed to obtain the image coordinates p of the laser points i , and then the coordinates of the laser points in the camera coordinate system are solved according to Equations (1) and (10) Using the laser point coordinates The dataset is optimized by RANSAC and then fitted by PCA to obtain the joint calibration results. Finally, the average distance from the laser points to the fitted line is 0.179 mm, and the spatial error of the laser points is 0.536 mm.

[0154] The feasibility and robustness of the proposed method are verified through simulation experiments and actual tests. The experimental results show that the method can effectively reduce the spatial error of the laser point coordinates and meet the required accuracy requirements, demonstrating the accuracy and practicality of the method.

[0155] In summary, through the above technical solutions, the present invention has successfully achieved high-precision joint calibration between the laser rangefinder and the camera, solved the problem of insufficient calibration accuracy in the prior art, and has important application value especially in the field of large antenna surface shape measurement.

[0156] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for calibrating a visual measurement system integrating laser ranging, characterized in that: The following steps are involved: S1. Establish the system coordinate model of the visual measurement system and the laser rangefinder, define the laser rangefinder coordinate system, the camera coordinate system and the world coordinate system, and clarify the spatial transformation relationship between the coordinate systems; S2. Calibrate the camera and obtain the camera's intrinsic and extrinsic matrix; S3. Obtain the image pixel coordinates of the laser point on the calibration plate; S4. Based on the camera intrinsic parameter matrix and extrinsic parameter matrix, the pixel coordinates of the laser point are converted into three-dimensional coordinates in the camera coordinate system; S5. Fitting the direction vector of the laser beam through the three-dimensional spatial distribution data of the laser point; S6. Based on the laser beam direction vector and the laser ranging value, calculate the external parameters of the laser rangefinder, including the rotation matrix and the translation matrix, and complete the calibration.

2. The method for calibrating a visual measurement system integrating laser ranging according to claim 1, characterized in that: In step S1, the establishment of the system coordinate model includes: The laser rangefinder coordinate system is defined as a three-dimensional coordinate system with the light-emitting point of the laser rangefinder as the origin and the direction of the laser beam as one of the coordinate axes; The camera coordinate system is defined as a three-dimensional coordinate system with the camera optical center as the origin and the optical axis direction as one of the coordinate axes; Define the world coordinate system as a three-dimensional coordinate system with the fixed point of the calibration plate as the origin; Establish the transformation relationship between the laser ranging coordinate system and the camera coordinate system, including the rotation matrix and the translation matrix, and define the transformation relationship between the camera coordinate system and the world coordinate system.

3. The method for calibrating a visual measurement system integrating laser ranging according to claim 1, characterized in that: In step S2, the camera calibration adopts the checkerboard calibration method, and the intrinsic parameters and extrinsic parameters of the camera are calculated by collecting checkerboard images at different positions and angles. The intrinsic parameters include focal length, principal point position and pixel scale factor, and the extrinsic parameters include rotation matrix and translation matrix of the camera relative to the world coordinate system.

4. The method for calibrating a visual measurement system integrating laser ranging according to claim 1, characterized in that: The method for obtaining the image pixel coordinates of the laser point in step S3 includes: Grayscale the captured image to separate the background and laser point area; The laser points are extracted using the threshold segmentation method; The noise is eliminated by morphological processing to obtain the accurate pixel coordinates of the laser point image.

5. The method for calibrating a visual measurement system integrating laser ranging according to claim 1, characterized in that: The method for converting the pixel coordinates of the laser point into three-dimensional coordinates in the camera coordinate system in step S4 includes: Use the camera intrinsic parameter matrix to project the pixel coordinates of the laser point to the camera coordinate system; Using the known spatial position information of the calibration plate, the laser point is mapped from the camera coordinate system to the world coordinate system through the camera extrinsic matrix; Based on the coordinate transformation relationship of the laser rangefinder, the laser point is transformed from the world coordinate system back to the camera coordinate system.

6. The method for calibrating a visual measurement system integrating laser ranging according to claim 1, characterized in that: The fitting of the laser beam direction vector in step S5 adopts the principal component analysis method, which specifically includes: Calculate the center point of the three-dimensional distribution of the laser points as the mean point; Calculate the covariance matrix of the laser point distribution; The covariance matrix is ​​subjected to eigenvalue decomposition, and the eigenvector corresponding to the maximum eigenvalue is extracted as the direction vector of the laser beam.

7. The method for calibrating a visual measurement system integrating laser ranging according to claim 6, characterized in that: In the laser beam direction vector fitting process in step S5, the laser point data is optimized by using a random sampling consistency algorithm, including: Randomly select two points from the laser point data set to fit a straight line; Calculate the distance from other points to the line, and take the points less than the set threshold as the inner points; Repeat sampling and select the model with the largest number of inliers as the fitting result; The best model internal point set is used to re-fit and determine the final laser beam direction vector.

8. The method for calibrating a visual measurement system integrating laser ranging according to claim 1, characterized in that: In step S6, the method for calculating the external parameters of the laser rangefinder based on the laser beam direction vector and the laser ranging value includes: According to the laser beam direction vector, calculate the rotation matrix of the laser rangefinder relative to the camera; Combined with the laser rangefinder value, the translation matrix of the laser rangefinder relative to the camera is calculated.

9. The method for calibrating a visual measurement system integrating laser ranging according to claim 1, characterized in that: After step S6 is completed, the calibration result is verified by an error evaluation method, including: Calculate the angle between the laser beam fitting direction vector and the true direction vector to evaluate the angle error; Calculate the Euclidean distance between the three-dimensional coordinates of the fitted laser point and the true coordinates to evaluate the position error; Through multiple data collection experiments, the error model is established and the calibration accuracy is analyzed.

10. A calibration device for a visual measurement system integrated with laser ranging, according to a calibration method for a visual measurement system integrated with laser ranging according to any one of claims 1 to 9, characterized in that: include: Laser rangefinder, used to generate a laser beam and measure the distance value of a target point; A camera for acquiring an image including the laser point; Calibration board, used to provide spatial reference information for camera and laser rangefinder; A data processing module is connected to the laser rangefinder and the camera for processing image data and extracting the pixel coordinates of the laser point; The operation module is connected to the data processing module and is used to calculate the external parameters of the laser rangefinder based on the laser point pixel coordinates, the laser distance measurement value and the coordinate model.

Images