A high-speed measurement system calibration error compensation method

Abstract

This invention provides a calibration error compensation method for a high-speed measurement system, relating to the fields of photoelectric non-contact, high-precision, and high-dynamic measurement technology. It solves the problems of error compensation for linear images not being parallel to the vertical direction of the image plane and the modeling of point-line imaging models. First, the centerline of the acquired linear image is fitted. Then, a multi-objective optimization model is established for the centerline, optimizing it into a straight line parallel to the vertical direction of the image plane to compensate for the centerline positioning algorithm error. Next, based on the relationship between the luminescent target and its projection onto the imaging plane as a straight line, multiple rays are established. A spatial straight line parallel to the centerline of the linear image is randomly selected from these rays, and a plane equation is established with the luminescent point. Finally, the plane equations of multiple luminescent targets are solved simultaneously to obtain the calibration parameters. This invention provides a calibration error compensation method for a high-speed measurement system that compensates for identification and positioning errors while simultaneously achieving calibration, with high calibration accuracy and good environmental robustness.

Description

Technical Field

[0001] This invention relates to a calibration error compensation method for a high-speed measurement system, and pertains to the field of photoelectric non-contact, high-precision, and high-dynamic measurement technology. Background Technology

[0002] The calibration error compensation method for a high-speed measurement system is based on a 3D pose measurement system using three one-dimensional linear scan cameras with integrated cylindrical-spherical lenses. The high-speed measurement system employs three one-dimensional linear scan cameras to capture images of a luminous target, which are then imaged as linear lines through optical lenses. The calculation unit solves for the linear equations of each line image formed by the one-dimensional linear scan cameras. These three linear equations, along with the luminous target, form three planes. By simultaneously solving the equations of these three planes, the spatial coordinates of the luminous target can be calculated. A key challenge in the calibration error compensation method for this high-speed measurement system is the accuracy of the calibration process, specifically the accuracy of the camera model parameter calculations. This accuracy is crucial for achieving 3D position measurement and improving overall precision.

[0003] Ideally, the three one-dimensional linear scan cameras in a high-speed measurement system should produce images that are straight lines parallel to the vertical direction of the image plane. However, in reality, most of the images identified have a slight tilt angle relative to the vertical direction of the image plane. This is mainly due to errors in the centerline positioning algorithm. Currently available algorithm models ignore the influence of this error, which presents three main challenges:

[0004] 1. It is difficult to accurately describe the model of how a one-dimensional linear scan camera combined with a cylindrical or spherical optical lens transforms a circular beam into a line beam. The technical documents currently available simplify the one-dimensional linear scan camera into an ideal pinhole model, and then establish a spatial solution model based on the imaging principle of an area scan camera. The algorithm is mature and simple, but this simplified model only retains the value (u, 0) or (v, 0) in one direction on the image plane, and directly ignores the model that satisfies parallel beam imaging in the v direction, which inevitably introduces errors.

[0005] 2. When identifying and locating the center line of the linear image in the image plane, the extreme case where the center line is not parallel to the vertical direction of the image plane is not considered. When the line resolution of the one-dimensional linear scan camera is greater than 1, theoretically the linear image in the image plane should be a straight line parallel to the vertical direction of the image plane. However, in reality, most of the identified images are not parallel to the vertical direction of the image plane. Therefore, a compensation algorithm for the identification and location needs to be developed.

[0006] 3. Solving the linear equation of the linear image and the planar mathematical model formed by the luminous target is difficult. The linear equation of the linear image is based on the image plane coordinate system, while the luminous target is based on the world coordinate system. How to establish the relationship between the two-dimensional image plane coordinate system and the world coordinate system of the high-speed measurement system is a technical problem that urgently needs to be solved.

[0007] The above three technical challenges limit the direct use of existing calibration and error compensation algorithms. Especially in practical engineering applications, high-speed measurement systems are severely affected by the on-site environment, leading to significant noise interference in imaging. The center line extracted from the center position of the fringe image cannot be parallel to the vertical direction of the image plane. If a one-dimensional linear array camera is simplified to an ideal pinhole model, ignoring positioning algorithm errors and the relationship between point and line imaging models, large errors will inevitably occur, and even measurement errors may result. Summary of the Invention

[0008] This invention is designed to address the shortcomings of the existing technology and provides a calibration error compensation method for a high-speed measurement system. Its purpose is to solve the error compensation problem of linear images not being parallel to the vertical direction of the image plane in a high-speed measurement system, as well as the drawback of simplifying the point-line imaging model to a hole-making imaging model when modeling.

[0009] The objective of this invention is achieved through the following technical solution:

[0010] The calibration error compensation method for this high-speed measurement system is characterized by the following steps:

[0011] Step A, the high-speed measurement system (1) acquires real-time data from the luminous target on the calibration device (2) P i Given two-dimensional images f1, f2, and f3 (i = 1, 2, ..., 20), use image processing algorithms to identify the linear images in images f1, f2, and f3 respectively, and fit the center lines l1, l2, and l3.

[0012] Step B: Establish an optimized model with corrected centerlines l1, l2, l3 as l'1, l'2, l'3 to compensate for the centerline positioning algorithm error.

[0013] B-1) Define a one-dimensional linear array camera image coordinate system I(u,v), where the direction of the image coordinate system u is horizontal to the right, the direction of v is vertically upward perpendicular to u, and the origin of the coordinate system is the center point of the image.

[0014] B-2) For the one-dimensional linear scan cameras on both sides, u = u in the image coordinate system. x To correct the centerlines l1 and l3, the following correction model is established:

[0015]

[0016] Using min(U) ij )≤u ix ≤max(U ij ) for u=u x Apply constraints;

[0017] B-3) For the intermediate one-dimensional linear scan camera, v = v in the image coordinate system x To correct the centerline l2, the following correction model is established:

[0018]

[0019] Using min(V) ij )≤v ix ≤max(V ij For v = v x Apply constraints;

[0020] B-4) The multi-objective solution model is constructed for the centerline correction of the three linear array cameras as follows:

[0021]

[0022] st min(U 1j )≤u 1x ≤max(U 1j )

[0023] min(U 3j )≤u 3x ≤max(U 3j )

[0024] min(V 2j )≤v 2x ≤max(V 2j )

[0025] The three objective functions λ1, λ2, and λ3 are used with equal weights, that is:

[0026] Step C: Construct a single-camera imaging model and solve for the luminous target point P. i The equations of the planes with respect to the centerlines l'1, l'2, l'3 are π1, π2, π3:

[0027] C-1) Define a one-dimensional linear scan camera and define the camera coordinate system C(x) c ,y c ,z c Camera coordinates x c Parallel to the right horizontally from u, camera coordinate y c Parallel to v and vertically upward, camera coordinate z c With x c and y c The right-hand rule applies, and the origin of the camera coordinate system is at the center of the camera.

[0028] C-2) For the two-sided one-dimensional linear array cameras, the pinhole imaging principle is satisfied in the direction of image coordinate system u, and the parallel beam imaging principle is satisfied in the direction of image coordinate system v. Based on the projection relationship, P is established. i Its relationship with its projection point u on the imaging plane;

[0029] C-3) For the intermediate one-dimensional linear scan camera, the pinhole imaging principle is satisfied in the direction of the image coordinate system v, and the parallel beam imaging principle is satisfied in the direction of the image coordinate system v. Based on the projection relationship, P is established. i Its relationship with its projection point v on the imaging plane:

[0030] C-4) For a one-dimensional linear array camera (3), take a point (u) on its imaging l'1. 1x v1), based on the above steps C-2), establish P i Projection point (u) on the imaging plane 1x The equation of the line L1 for v1 is:

[0031] C-5) For a one-dimensional linear array camera (3), take a point (u) on its imaging l'1. 1x v2), based on the above steps C-2), establish P i Projection point (u) on the imaging plane 1x The equation of the line L2 for v2) is:

[0032] C-6) Take any point T0(x) on line L1. 11 ,y 11 ,z 11 Solve for the point T0(x) 11 ,y 11 ,z 11 And parallel to the camera coordinate y c Given a line S1 on the x-axis, find the intersection point T1(x) of line S1 and line L2. 12 ,y 12 ,z 12 );

[0033] C-7) Move line L1 along the camera coordinate y c The equation of the line L'1 is obtained by shifting the axis downwards by an angle α.

[0034] Angle α must satisfy the following requirements:

[0035]

[0036] In the formula, angle α is defined as the luminous target point P. i The angle between the line connecting the two endpoints of the center line of the two-dimensional linear image, d is the working distance of the target point from the camera, row is the row resolution of the linear camera, and du is the pixel size of the one-dimensional linear camera.

[0037] C-8) Solve for the relationship between line L'1 and camera coordinates y c The coordinates of the intersection point T of the axes i (x 1i ,y 1i ,z 1i );

[0038] C-9) based on T0(x 11 ,y 11 ,z 11 ), T1(x 12 ,y 12 ,z 12 ), T i (x 1i ,y 1i ,z 1i ) and P i At least four points are needed to establish the plane equation π1;

[0039] Similarly, based on the steps C-1) to C-9) above, establish the plane equations π2 and π3.

[0040] Step D: Combine the multi-point imaging models to solve for the system calibration parameters.

[0041] D-1) Collect at least 20 luminescent target points and establish plane equations according to the above steps;

[0042] D-2) Solve the system of multiple equations to obtain the calibration parameters.

[0043] The features and technical effects of the technical solution of this invention are as follows:

[0044] 1. The technical solution of this invention uses a one-dimensional line array camera with cylindrical and spherical optical lenses to transform a circular beam into a line beam and solve for calibration parameters using a point-line model. Instead of simply simplifying the one-dimensional line array camera into an ideal pinhole model, this invention solves the modeling problem of point-line imaging models, making the algorithm model more realistic and accurate, and the calibration precision higher.

[0045] 2. The technical solution of this invention establishes a multi-objective solution model to correct the center line of the image, which solves the problem of error compensation for linear images that are not parallel to the vertical direction of the image plane. This processing can remove the interference of positioning algorithm error, reduce the influence of various noises in the actual application environment, ensure the accuracy and precision of two-dimensional planar image extraction, and meet the high-precision measurement requirements of high-speed measurement systems.

[0046] 3. The technical solution of this invention proposes a planar equation fitting method based on the projection relationship between points and lines. This method establishes multiple rays based on the relationship between the luminescent target and its projection onto the imaging plane as a straight line. One spatial straight line parallel to the center line of the linear image is randomly selected from these rays. This straight line and the luminescent point establish a planar equation. This method accurately represents the point-line imaging model, making the algorithm design more reasonable. The points used to fit the planar equation are virtual points outside the image plane, expanding the distance between non-collinear points and improving the accuracy of the planar equation fitting, thus making the system measurement accuracy more precise.

[0047] 4. The technical solution of the present invention can perform high-precision calibration of high-speed measurement systems, and complete the calibration parameter solution while compensating for the error of the two-dimensional planar image positioning algorithm. It has good environmental adaptability, high accuracy, and wide applicability. It is suitable for long strip one-dimensional line scan cameras, and can also be extended to area scan cameras and line scan cameras with a line resolution of 1. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the method of the present invention;

[0049] Figure 2 This is a schematic diagram of the high-speed measurement system (1) in this embodiment of the invention acquiring the luminous target (2) on the calibration device in real time;

[0050] Figure 3 This is a schematic diagram of the installation positions of the three one-dimensional line array cameras (3), (4), and (5) included in the high-speed measurement system (1) in this embodiment of the invention;

[0051] Figure 4 In this embodiment of the invention, the center lines l1, l2, l3 and l'1, Schematic diagram of the positional relationship of l'3;

[0052] Figure 5 A schematic diagram of the spatial model for establishing the plane equation π1 in this embodiment of the invention;

[0053] Figure 6 This is a schematic diagram of the imaging model of the one-dimensional linear array camera in the linear array measurement device (1) of this invention when the row is 1. Detailed Implementation

[0054] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments:

[0055] See appendix Figure 1 As shown, the steps of the high-speed measurement system calibration error compensation method according to the embodiment of the present invention are as follows:

[0056] Step A, the high-speed measurement system (1) acquires real-time data from the luminous target on the calibration device (2) Pi Given two-dimensional images f1, f2, and f3 (i = 1, 2, ..., 20), use image processing algorithms to identify the linear images in images f1, f2, and f3 respectively, and fit the center lines l1, l2, and l3.

[0057] In this application example, the high-speed measurement system (1) includes three one-dimensional linear array cameras (3), (4), and (5) to acquire real-time data from the luminous target (2) P on the calibration device. i Two-dimensional images f1, f2, and f3 of (i = 1, 2, ..., 20) are shown in the diagram below. Figure 2 As shown.

[0058] For the acquired two-dimensional images f1, f2, and f3, center lines l1, l2, and l3 are fitted. In this application example, traditional image processing algorithms are mainly used to perform binarization, filtering, edge extraction, and Hough transform on the images to extract the center lines of the linear images.

[0059] Step B: Establish an optimized model with corrected centerlines l1, l2, l3 as l'1, l'2, l'3 to compensate for the centerline positioning algorithm error.

[0060] B-1) Define a one-dimensional linear array camera image coordinate system I(u,v), where the direction of u in the image coordinate system is horizontal to the right, and the direction of v is vertically upwards perpendicular to u. The origin of the coordinate system is the center point of the image. Figure 4 As shown;

[0061] B-2) For the one-dimensional linear scan cameras on both sides, u = u in the image coordinate system. x To correct the centerlines l1, l3 to l'1, l'3, the following corrected model is established:

[0062]

[0063] Using min(U) ij )≤u ix ≤max(U ij ) for u=u x Apply constraints;

[0064] B-3) For the intermediate one-dimensional linear scan camera, v = v in the image coordinate system x To correct the centerline l2 to l'2, the following correction model is established:

[0065]

[0066] Using min(V) ij )≤v ix ≤max(V ij For v = v x Apply constraints;

[0067] The relationship between l1 and l'1 and l2 and l'2 in steps B-2) and B-3) above is illustrated in the diagram below. Figure 4 As shown.

[0068] B-4) The multi-objective solution model is constructed for the centerline correction of the three linear array cameras as follows:

[0069]

[0070] st min(U 1j )≤u 1x ≤max(U 1j )

[0071] min(U 3j )≤u 3x ≤max(U 3j )

[0072] min(V 2j )≤v 2x ≤max(V 2j )

[0073] The three objective functions λ1, λ2, and λ3 are used with equal weights, that is: The lines l'1, l'2, and l'3 can then be solved.

[0074] Step C: Construct a single-camera imaging model and solve for the luminous target point P. i The plane equations π1, π2, π3 of the centerlines l'1, l'2, l'3 are as follows: The spatial model is... Figure 5 As shown:

[0075] C-1) Define a one-dimensional linear scan camera and define the camera coordinate system C(x) c ,y c ,z c Camera coordinates x c Parallel to the right horizontally from u, camera coordinate y c Parallel to v and vertically upward, camera coordinate z c With x c and y c The right-hand rule applies, and the origin of the camera coordinate system is at the center of the camera.

[0076] C-2) For the two-sided one-dimensional linear array cameras, the pinhole imaging principle is satisfied in the direction of image coordinate system u, and the parallel beam imaging principle is satisfied in the direction of image coordinate system v. Based on the projection relationship, P is established. i Its relationship with its projection point u on the imaging plane;

[0077] In this application example, P iThe world coordinates are P i (x i ,y i z i Its projection point on the imaging plane is:

[0078]

[0079] C-3) For the intermediate one-dimensional linear scan camera, the pinhole imaging principle is satisfied in the direction of the image coordinate system v, and the parallel beam imaging principle is satisfied in the direction of the image coordinate system v. Based on the projection relationship, P is established. i Its relationship with its projection point v on the imaging plane:

[0080] In this application example, P i The world coordinates are P i (x i ,y i z i Its projection point on the imaging plane is:

[0081] (1,v 2x )

[0082] (2,v 2x ) ...

[0084] (20,v 2x )

[0085] C-4) For a one-dimensional linear array camera (3), take a point (u) on its imaging l'1. 1x v1), based on the above steps C-2), establish P i Projection point (u) on the imaging plane 1x The equation of the line L1 for v1 is:

[0086] C-5) For a one-dimensional linear array camera (3), take a point (u) on its imaging l'1. 1x v2), based on the above steps C-2), establish P i Projection point (u) on the imaging plane 1x The equation of the line L2 for v2) is:

[0087] C-6) Take any point T0(x) on line L1. 11 ,y 11 ,z 11 Solve for the point T0(x) 11 ,y 11 ,z 11 And parallel to the camera coordinate y c Given a line S1 on the x-axis, find the intersection point T1(x) of line S1 and line L2. 12 ,y 12 ,z12 );

[0088] C-7) Move line L1 along the camera coordinate y c The equation of the line L'1 is obtained by shifting the axis downwards by an angle α.

[0089] Angle α must satisfy the following requirements:

[0090]

[0091] In the formula, angle α is defined as the luminous target point P. i The angle between the line connecting the two endpoints of the center line of the two-dimensional linear image, d is the working distance of the target point from the camera, row is the row resolution of the linear camera, and du is the pixel size of the one-dimensional linear camera.

[0092] In this application example, the high-speed measurement system has a measurement range of 5–30 meters, therefore a working distance d of 7–10 meters is selected for calibration. The value of du can be 3.5µm, 5µm, 7µm, 14µm, etc. In this application example, a one-dimensional linear array camera with a 5µm pixel size is selected. The angle α is set to α = 1°.

[0093] C-8) Solve for the relationship between line L'1 and camera coordinates y c The coordinates of the intersection point T of the axes i (x 1i ,y 1i ,z 1i );

[0094] C-9) based on T0(x 11 ,y 11 ,z 11 ), T1(x 12 ,y 12 ,z 12 ), T i (x 1i ,y 1i ,z 1i ) and P i At least four points are needed to establish the plane equation π1;

[0095] Similarly, based on the steps C-1) to C-9) above, establish the plane equations π2 and π3.

[0096] Specifically: When the row resolution of a one-dimensional linear scan camera is 1, the spatial model is transformed as follows: Figure 6 As shown. If the image on a one-dimensional linear array camera is a single point, then the point-line imaging model simplifies to a point-to-point imaging model, and the plane equation simplifies to a straight line equation.

[0097] Step D: Combine the multi-point imaging models to solve for the system calibration parameters.

[0098] D-1) Collect at least 20 luminescent target points and establish plane equations according to the above steps;

[0099] D-2) Solve the system of multiple equations to obtain the calibration parameters.

Claims

1. A method for compensating calibration errors in a high-speed measurement system, characterized in that: The steps of this method are as follows: Step 1: The high-speed measurement system (1) collects data in real time from the luminous target on the calibration device (2). P i ( i Two-dimensional image of (=1,2...20) f 1. f 2 and f 3. Image processing algorithms are used to identify the images respectively. f 1. f 2 and f The center line of the linear image in 3 is fitted. l 1, l 2, l 3; Imaging is achieved using three one-dimensional line-scan cameras (3), (4), and (5) with an integrated cylindrical-spherical optical lens. The three one-dimensional line-scan cameras respectively transform the circular beam of light from the luminous target into a line beam. f 1 and f 3 represents images captured by the one-dimensional linear scan cameras on both sides. f 2 is the image captured by the central one-dimensional linear camera; Step 2: Establish the correction centerline l 1, l 2, l 3 is , , An optimized model is used to compensate for errors in the centerline positioning algorithm. Step 3: Construct a single-camera imaging model and solve for the luminous target points. P i With the center line , , plane equation π 1, π 2, π 3, include: 1) Define a one-dimensional linear scan camera and define the camera coordinate system. C ( x c , y c , z c Camera coordinates x c Parallel to u Horizontal to the right, camera coordinates y c Parallel and v Vertically upward, camera coordinates z c and x c and y c The right-hand rule applies, and the origin of the camera coordinate system is at the center of the camera. 2) For the one-dimensional linear scan cameras on both sides, in the image coordinate system u The direction satisfies the principle of pinhole imaging, and the image coordinate system v The direction satisfies the principle of parallel beam imaging, and the projection relationship is established. P i Its relationship with its projection point on the imaging plane; 3) For the intermediate one-dimensional linear scan camera, in the image coordinate system v The direction satisfies the principle of pinhole imaging, and the image coordinate system v The direction satisfies the principle of parallel beam imaging, and the projection relationship is established. P i Its relationship with its projection point on the imaging plane; 4) Overluminescent target P i ( i =1,2...20) respectively with the corrected centerline , , Establish plane equations π 1, π 2, π 3; Establish plane equations π 1, π 2, π Step 3 includes: For a one-dimensional linear scan camera (3), take its image The previous point, establish P i The equation of the straight line between the projection point of this point on the imaging plane and the projection point of this point L 1; For a one-dimensional linear scan camera (3), take its image The previous point, establish P i The equation of the straight line between the projection point of this point on the imaging plane and the projection point of this point L 2; In a straight line L 1. Take a point from the top T 0( x 11 , y 11 , z 11 Solve for the point passing through. T 0( x 11 , y 11 , z 11 And parallel to the camera coordinate axis y c straight line S 1. Solve for the straight line S 1 and the line L The intersection of 2 T 1( x 12 , y 12 , z 12 ); straight line L 1. Along the camera coordinates y c Translate downward along the axis α Angles yield the equation of a straight line ; Solve for the straight line With camera coordinates y c coordinates of the intersection of the axes T i ( x 1i , y 1i , z 1i ); based on T 0( x 11 , y 11 , z 11 ), T 1( x 12 , y 12 , z 12 ), T i ( x 1i , y 1i , z 1i )and P i At least four points are needed to establish the plane equation π 1; Similarly, establish the plane equations π 2, π 3; Step 4: Combine the multi-point imaging models to complete the solution of system calibration parameters.

2. The calibration error compensation method for a high-speed measurement system as described in claim 1, characterized in that: The light-emitting targets on the calibration device in step one are light-emitting LEDs, and the number of them is no less than 20.

3. The calibration error compensation method for a high-speed measurement system as described in claim 1, characterized in that: The α The angle must meet the following requirements: ； In the formula α Angle is defined as the luminous target point. P i The angle between the line connecting the two endpoints of the center line of the two-dimensional linear image and the line connecting the two endpoints of the center line. d The target point is the working distance from the camera. row For line scan camera row resolution, du This refers to the pixel size of a one-dimensional line scan camera.

Images