Four-wheel alignment single-camera calibration method and system

Abstract

The application provides a four-wheel positioning single-camera calibration method and system, and belongs to the field of automobile detection and diagnosis equipment.The method comprises camera focal length adjustment, target image acquisition, distortion correction of target image pixels by using a camera distortion model, camera intrinsic parameter calibration by using the corrected target pixels, and calculation of intrinsic parameters to complete calibration.The measured intrinsic parameters of the application have high accuracy, and the precision completely meets the four-wheel positioning requirement.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automobile detection and diagnosis equipment, in particular to a four-wheel positioning single camera calibration method and system. BACKGROUND

[0002] In the automobile 3D four-wheel positioning, double cameras or multiple cameras are used to detect the tire position parameters by using image processing technology. Among them, the camera needs to be calibrated after new installation or replacement to ensure the accuracy of the positioning parameters. The camera calibration method is to obtain the target picture by shooting the target with the camera, extract the pixel point coordinates in the target picture by using the algorithm, and correspond the pixel point coordinates with the world coordinates of the target coordinate system, so as to establish the conversion relationship between them and solve the camera calibration parameters.

[0003] In the prior art, when a circular target is used to solve the camera calibration parameters, four feature points not in a straight line are randomly selected from the target feature points to solve the homography matrix and then solve the camera intrinsic parameters. However, because the target image will be radially distorted and tangentially distorted in the single camera calibration process, if the collected data is directly used for intrinsic parameter calibration, the obtained intrinsic parameter is inaccurate, which further affects the subsequent four-wheel positioning.

[0004] Therefore, there is an urgent need for a single camera calibration method that can improve the accuracy of intrinsic parameters and meet the accuracy requirements of four-wheel positioning. SUMMARY

[0005] To solve the above technical problems, the present application provides a four-wheel positioning single camera calibration method and system. The method corrects the radial distortion and tangential distortion of the collected image by using a distortion model, and then performs intrinsic parameter calibration. The measured intrinsic parameter has high accuracy and fully meets the four-wheel positioning requirements.

[0006] To solve the above technical problems, the present application adopts the following technical solutions:

[0007] A four-wheel positioning single camera calibration method, comprising the following steps:

[0008] Step (1): Camera focal length adjustment: place the camera lens at the center of the focal length adjustment plate, adjust the camera focal length and fix it;

[0009] Step (2): Target image acquisition:

[0010] Install the calibration target on the calibration support, place the camera lens at the center of the calibration target, turn the calibration target left and right and pitch, and use the camera to calibrate and collect target images at multiple angles;

[0011] Step (3): Distortion correction of the pixel points of the target image by using the camera distortion model:

[0012] The pixel points of the non-origin and non-center points of the target image will have radial distortion and tangential distortion, and the coordinates of the pixel points after distortion are p'(x d ,y d ), p(x,y) is the estimated coordinates of the pixel points after correction by the distortion model, and is the estimated value of the ideal coordinates, so the distortion model is:

[0013] The horizontal axis radial distortion d rx in the projection process = x(1+k1r 2 +k2r 4 +k3r 6 );

[0014] The vertical axis radial distortion d ry in the projection process = y(1+k1r 2 +k2r 4 +k3r 6 );

[0015] The horizontal axis tangential distortion d tx in the projection process = 2p1xy+p2(r 2 +2x 2 );

[0016] The vertical axis tangential distortion d ty in the projection process = 2p2xy+p1(r 2 +2y 2 );

[0017] The distortion coordinates in the projection process are

[0018] where r 2 =x 2 +y 2 , k1 is the quadratic radial coefficient, k2 is the fourth radial coefficient, k3 is the sixth radial coefficient, p1 is the horizontal axis tangential coefficient, and p2 is the vertical axis tangential coefficient.

[0019] Step (4): Using the corrected target pixel points to calibrate the camera intrinsic parameters, calculating the intrinsic parameters, completing the calibration, specifically:

[0020] Let the target center coordinates be (X C ,Y C ,Z C ), and the image center pixel coordinates (u,v) on the target are obtained by shooting photos during single camera calibration, and the camera internal characteristic parameters are expressed by a three-dimensional matrix expression: s is the scale factor, then the data in step (3) is brought into the following formula:

[0021]

[0022] The internal characteristic parameters of each camera are calculated, including the normalized focal length fx of the X axis, the normalized focal length fy of the Y axis, and the image center coordinate parameter (u0, v0).

[0023] In the further technical solution, the specific method for adjusting the focal length of the camera in step (1) is as follows: the lens of the camera is placed at the center of the focal length adjustment plate, the camera is used to capture the focal length adjustment target at a distance of 1.7 meters and 2.5 meters respectively, and the best focal length of the camera for viewing the near target and the far target is obtained.

[0024] In the further technical solution, the specific method for adjusting the focal length of the camera in step (1) is as follows: the lens of the camera is placed at the center of the focal length adjustment plate, the camera is used to capture the focal length adjustment target at a distance of 1.7 meters and 2.5 meters respectively, and the best focal length of the camera for viewing the near target and the far target is obtained.

[0025] 1) 7 times of horizontal upward adjustment: 5°, 10°, 15°, 20°, 25°, 30° and 35° of horizontal upward adjustment;

[0026] 2) 7 times of horizontal downward adjustment: 5°, 10°, 15°, 20°, 25°, 30° and 35° of horizontal downward adjustment;

[0027] 3) 7 times of vertical left turning adjustment: 5°, 10°, 15°, 20°, 25°, 30° and 35° of vertical left turning adjustment;

[0028] 4) 7 times of vertical right turning adjustment: 5°, 10°, 15°, 20°, 25°, 30° and 35° of vertical right turning adjustment;

[0029] In the further technical solution, the calibration target is a 7*7 circular dot target or a circular dot target with an array specification of 7*7 or above.

[0030] The four-wheel positioning single-camera calibration system provided by the application comprises a calibration target, a camera and software.

[0031] Advantages

[0032] Compared with the prior art, the application has the following advantages:

[0033] 1. The target in the present application can adopt a circular array of 7*7 specifications or above, more feature points are extracted from each image, and 29 (and above) images can be collected in calibration, so that the calibration data is more sufficient, and the calibration accuracy is high.

[0034] 2. The present application can effectively reduce the influence of camera lens radial distortion and tangential distortion on the calibration result by rotating the target along the horizontal and vertical axes and collecting images, so that the error of the calibration data caused by the camera lens distortion is greatly reduced, and then the data is further corrected through the distortion model to improve the accuracy of the intrinsic parameter parameters measured in the intrinsic parameter calibration, and meet the four-wheel positioning requirements.

[0035] 3. The calibration method of the present application can be used for the calibration of the left and right cameras of the four-wheel alignment instrument, the calibration method is easy to operate, and the requirement for maintenance personnel is low, without the need for complex training of maintenance personnel, improving the calibration efficiency, shortening the four-wheel positioning time, and reducing the cost of four-wheel positioning. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The flow chart of the four-wheel positioning single camera calibration method in the embodiment of the present application;

[0037] Figure 2 The schematic diagram of camera distortion in the embodiment of the present application;

[0038] Fig. 3(a) is a schematic diagram of front calibration of the calibration target in the embodiment of the present application;

[0039] Fig. 3(b) is a schematic diagram of horizontal upward calibration of the calibration target in the embodiment of the present application;

[0040] Fig. 3(c) is a schematic diagram of horizontal downward calibration of the calibration target in the embodiment of the present application;

[0041] Fig. 3(d) is a schematic diagram of vertical left turning calibration of the calibration target in the embodiment of the present application;

[0042] Fig. 3(e) is a schematic diagram of vertical right turning calibration of the calibration target in the embodiment of the present application;

[0043] Figure 4 The calibration result diagram in the embodiment of the present application;

[0044] Figure 5 The calibration pixel error result in the embodiment of the present application. DETAILED DESCRIPTION

[0045] For the convenience of understanding the present application, the following will make a more comprehensive and detailed description of the present application in conjunction with the accompanying drawings and preferred embodiments, but the scope of protection of the present application is not limited to the following specific embodiments. The embodiments of the present application are described in detail below in conjunction with the drawings, but the present application can be implemented in various different ways limited and covered by the claims.

[0046] Embodiment 1

[0047] A four-wheel positioning single-camera calibration method, comprising the following steps:

[0048] Step (1): Camera focal length adjustment: place the camera lens at the center of the focal length adjustment plate, adjust the camera focal length and fix it; the specific method for adjusting the camera focal length is: place the camera lens at the center of the focal length adjustment plate, use the camera to shoot the focal length adjustment target at a distance of 1.7 meters and 2.5 meters respectively, obtain the best focal length for the camera to view the near target and the far target, and when the camera shoots the focal length adjustment plate picture, the scale of the four edges needs to be clearly seen, and the scale with the largest value needs to be clearly seen.

[0049] Step (2): Target image acquisition:

[0050] Install the calibration target on the calibration support, place the camera lens at the center of the calibration target, turn the calibration target left and right and pitch, and use the camera to calibrate and acquire target images at multiple angles;

[0051] Step (3): Distortion correction of pixel points of the target image using the camera distortion model:

[0052] The pixel points of the non-origin and non-center points of the target image will occur radial distortion and tangential distortion, as shown in Figure 2 , let the coordinates of the pixel points after distortion be p'(x d ,y d ), p(x,y) is the estimated coordinates of the pixel points after correction by the distortion model, and is the estimated value of the ideal coordinates, the change of the lens distortion reflected on the image is that each distorted pixel point has changed in distance compared with the ideal pixel point:

[0053] The horizontal axis radial distortion d rx =x(1+k1r 2 +k2r 4 +k3r 6 ) in the projection process;

[0054] The vertical axis radial distortion d ry =y(1+k1r 2 +k2r 4 +k3r 6 ) in the projection process;

[0055] Horizontal tangential distortion d during projection tx =2p1xy+p2(r 2 +2x 2 );

[0056] Vertical tangential distortion d during projection ty =2p2xy+p1(r 2 +2y 2 );

[0057] Therefore, distortion correction involves adding the above distortion models together to obtain the final distortion model:

[0058] The distorted coordinates during the projection process are

[0059] Where, r 2 =x 2 +y 2 k1 is the second radial coefficient, k2 is the fourth radial coefficient, k3 is the sixth radial coefficient, p1 is the transverse tangential coefficient, and p2 is the longitudinal tangential coefficient.

[0060] Step (4): Use the calibrated target pixels to perform camera intrinsic parameter calibration, calculate the intrinsic parameter parameters, and complete the calibration. Single-camera intrinsic parameter calibration is achieved by calculating the mapping relationship between the three-dimensional coordinates of feature points on the calibration target and their corresponding pixel coordinates to solve for the camera's intrinsic parameters. Specifically:

[0061] Let the coordinates of the target center be (X). C ,Y C Z C The image center pixel coordinates (u,v) on the target are obtained by taking photos during single-camera calibration. The camera's internal characteristic parameters are expressed using a three-dimensional matrix. If s is a scaling factor, then substitute the data from step (3) into the following formula:

[0062]

[0063] The internal characteristic parameters of each camera are calculated as follows: normalized focal length fx on the X-axis, normalized focal length fy on the Y-axis, and image center coordinate parameters (u0, v0).

[0064] like Figures 3(a)-3(e) As shown, taking a 7*7 circular target as an example, the specific calibration method in step (2) is as follows: First, calibrate the front of the target once, then rotate the target left, right and up respectively, and collect an image every 5° of rotation, with no less than 7 sets of data in each direction, for calibration. The specific method is as follows:

[0065] 1) Horizontal tilt calibration 7 times: in the following order: horizontal tilt 5°, horizontal tilt 10°, horizontal tilt 15°, horizontal tilt 20°, horizontal tilt 25°, horizontal tilt 30°, horizontal tilt 35°;

[0066] 2) Horizontal downward calibration 7 times: in the following order: horizontal downward 5°, horizontal downward 10°, horizontal downward 15°, horizontal downward 20°, horizontal downward 25°, horizontal downward 30°, and horizontal downward 35°;

[0067] 3) Vertical left turn marking 7 times: in the following order: vertical left turn 5°, vertical left turn 10°, vertical left turn 15°, vertical left turn 20°, vertical left turn 25°, vertical left turn 30°, vertical left turn 35°;

[0068] 4) Vertical right turn calibration 7 times: in the following order: vertical right turn 5°, vertical right turn 10°, vertical right turn 15°, vertical right turn 20°, vertical right turn 25°, vertical right turn 30°, vertical right turn 35°;

[0069] Due to differences in parameters such as wide-angle, cameras need to be calibrated for targets at multiple angles and positions. Using the above calibration method, a total of 29 sets of target photos from different angles were collected. When calculating the internal parameters, the calibration data was sufficient and the accuracy was high.

[0070] Example 2

[0071] A four-wheel alignment single-camera calibration system includes a calibration target, a camera, and software, wherein the software is used to execute the four-wheel alignment single-camera calibration method in Embodiment 1.

[0072] The above calibration method was used to calibrate the two cameras of a certain four-wheel alignment machine. The experimental results obtained after calibrating the intrinsic parameters of the two cameras are as follows: Figure 4 As shown, the calibration results were used to calculate the error, and the results are as follows. Figure 5 As shown, from Figure 4 As can be seen, the calculation errors of all parameters calibrated for both cameras are less than 0.4 pixels, demonstrating high accuracy and verifying the feasibility of using the method of this invention for camera calibration. The calibration method of this invention considers the correction of radial and tangential distortion, and its accuracy fully meets the requirements of four-wheel alignment.

Claims

1. A four-wheel alignment single camera calibration method, characterized by: It comprises the following steps: Step (1): camera focal length adjustment: place the camera lens at the center of the focal length adjustment board, adjust the camera focal length and fix it; Step (2): target image acquisition: Install the calibration target on the calibration support, place the camera lens at the center of the calibration target, and turn the calibration target left and right and pitch, respectively, and use the camera to calibrate and acquire target images at multiple angles; Step (3): use the camera distortion model to correct the distortion of the target image pixels: The pixel points of non-origin and non-center points of the target image will have radial distortion and tangential distortion. Let the coordinates of the pixel points after distortion be p'(x d ,y d ), and p(x,y) be the estimated coordinates of the pixel points after correction by the distortion model. The distortion model is: Horizontal axis radial distortion d in the projection process rx = x (1 + k1r 2 + k2r 4 + k3r 6 ); Longitudinal axis radial distortion d in the projection process ry = y (1 + k1r 2 + k2r 4 + k3r 6 ); Horizontal tangential distortion d in the projection process tx = 2p1xy + p2(r 2 + 2x 2 ); Longitudinal tangential distortion d in the projection process ty = 2p2xy + p1(r 2 + 2y 2 ); The distorted coordinates in the projection process are wherein r 2 = x 2 + y 2 , k1 is a quadratic radial coefficient, k2 is a quartic radial coefficient, k3 is a sextic radial coefficient, p1 is a tangential coefficient of the transverse axis, and p2 is a tangential coefficient of the longitudinal axis. Step (4): use the corrected target pixels to calibrate the camera intrinsic parameters, calculate the intrinsic parameters, and complete the calibration.

2. The four-wheel alignment single-camera calibration method of claim 1, wherein: In step (1), the specific method for adjusting the camera focal length is: place the camera lens at the center of the focal length adjustment board, use the camera to shoot the focal length adjustment target at a distance of 1.7 meters and 2.5 meters, respectively, to obtain the best focal length for the camera to view the near and far targets. When the camera shoots the focal length adjustment board, the scale of the four edges needs to be clearly visible, and the largest scale value needs to be clearly visible.

3. The four-wheel alignment single-camera calibration method of claim 1, wherein: In step (2), the specific method for calibration is: first calibrate the front of the calibration target, then turn the calibration target left and right and pitch, acquire an image every 5°, and calibrate at least 7 groups of data in each direction.

4. The four-wheel alignment single-camera calibration method of claim 1 or 3, wherein: The specific method for calibration of the calibration target turning left and right and pitch: 1) Calibrate 7 times horizontally and upward: 5° horizontally and upward, 10° horizontally and upward, 15° horizontally and upward, 20° horizontally and upward, 25° horizontally and upward, 30° horizontally and upward, 35° horizontally and upward; 2) Calibrate 7 times horizontally and downward: 5° horizontally and downward, 10° horizontally and downward, 15° horizontally and downward, 20° horizontally and downward, 25° horizontally and downward, 30° horizontally and downward, 35° horizontally and downward; 3) Calibrate 7 times vertically and left: 5° vertically and left, 10° vertically and left, 15° vertically and left, 20° vertically and left, 25° vertically and left, 30° vertically and left, 35° vertically and left; 4) Calibrate 7 times vertically and right: 5° vertically and right, 10° vertically and right, 15° vertically and right, 20° vertically and right, 25° vertically and right, 30° vertically and right, 35° vertically and right.

5. The four-wheel alignment single-camera calibration method of claim 1, wherein: The calibration target uses a 7*7 dot target and above array specification dot target.

6. The four-wheel alignment single-camera calibration method of claim 1, wherein: The specific method for intrinsic parameter calibration in step (4) is: Let the target center coordinates be (X C ,Y C ,Z C ), the image center pixel coordinates on the target be (u,v) obtained by taking a photo during single camera calibration, and the camera internal characteristic parameters be expressed in a three-dimensional matrix expression: s is a scale factor, then the data in step (3) is brought into the following formula: Calculate the internal characteristic parameters of each camera: X-axis normalized focal length fx, Y-axis normalized focal length fy, and image center coordinate parameter (u0, v0).

7. A four-wheel alignment single camera calibration system, comprising: It comprises a calibration target, a camera and software, and the software is used to execute the method of any one of claims 1-6.

Images