Method for calibrating vehicle headlights

Abstract

The present invention relates to a method (1) for calibrating vehicle headlights, in which a misalignment of a predetermined light distribution is determined by evaluating images from at least one vehicle camera, characterized in that in a first step (2) at least three images are taken, namely a first image (3) including a calibration pattern (4), a second image (5) without the calibration pattern (4), and a third image (6) including an inverted calibration pattern (4'), and in a second step (7) the second image (5) is subtracted from the first and third images (3, 6) to clean up the images (3, 6) from the existing scene.

Description

[Technical Field]

[0001] The present invention relates to a method for calibrating vehicle headlights as defined in more detail in the preamble of claim 1. [Background technology]

[0002] Basically, methods for calibrating vehicle headlights are known from the prior art. This type of method is necessary because vehicle headlights are often misaligned. Even if a vehicle is equipped with an automatic lighting range control, over time, mechanical stress, thermal expansion or contraction, and step loss of the attached stepping motor occur. All of these factors cause the headlights to gradually misalign.

[0003] Vertical misalignment, in particular, can dazzle oncoming traffic and reduce your own field of vision, creating a safety issue. Therefore, it is advisable to inspect headlight position periodically.

[0004] Patent document 1 thus discloses a method for generating a three-dimensional depth information map of an environment. In this method, a light pattern is projected onto the environment using a projector and captured by a camera. The camera images can then be evaluated, with the respective positions of feature points on corresponding epipolar lines being identified, and depth information for the three-dimensional map being obtained by determining the positional displacements of the feature points on these lines.

[0005] Another prior art is described in Patent Document 2. This prior art describes a camera device for a vehicle that can perform 3D ambient environment detection. For this purpose, at least two camera modules with at least partially overlapping detection areas are required. Via a control unit, an evaluation unit, and a spotlight projector, 3D ambient environment detection can be performed using a "pseudo-noise pattern."

[0006] US Patent No. 5,629,999 describes a method for adjusting the headlights of a vehicle. To this end, the individual lighting units of the headlights are controlled in a time-shifted manner to illuminate a scene. The scene is photographed over a predetermined period of time using a camera included in the vehicle. The resulting luminance distribution pattern obtained by the time-shifted control of multiple lighting units allows for the calculation of deviations from a reference pattern. The headlights can then be adjusted based on the deviations. For further prior art, see US Patent Nos. 5,629,999; ... and 5,629,999.

[0007] Stepper motors are used to actively adjust headlights, for example for illumination range control or cornering light functions. These stepper motors can adjust the lighting modules in the headlights by a desired angle. Approaches also exist for automatically and on-the-fly adjusting the headlight position. For example, the light distribution of the headlights in the area ahead of the vehicle can be detected and evaluated using a driver assistance camera. For this type of evaluation, prominent points with clear boundaries (cut-off lines), such as the H0V0 point, are often used. An attempt is made to identify these points in the image. In another variant, a specially designed light distribution for calibration can also be emitted in appropriate situations, such as during startup or during sleep phases.

[0008] Next, based on the distance from each pixel in the camera image to the calibrated reference point, the vertical and horizontal angular differences are estimated and compared with the current target adjustment. This allows the vertical and horizontal misalignment to be compensated. However, difficulties arise in uncontrolled or uncontrollable environmental conditions. These environmental conditions may include light characteristics, as well as the structure, shape, and position of the illuminated surface. Different vehicles may also have different lighting systems. In this case, differences in color and brightness distribution may result in individual pixel errors or inaccuracies in the light path, which may cause blurring and color shifts. Evaluation algorithms may be based, inter alia, on edge and maximum detection, however, this may lead to loose calibration with inaccurate and variable feature extraction. In particular, it may not be possible to robustly maintain the ECE-compliant safety-relevant accuracy of 0.1%. The introduction of high-resolution lighting systems into vehicles, for example based on LCD, DMD or μLED technology, also makes it necessary to project increasingly complex light distributions. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] DE102020007613A1 [Patent Document 2] DE102017222708A1 [Patent Document 3] DE102016118801A1 [Patent Document 4] DE102012007908A1 [Patent Document 5] DE102016006391A1 [Patent Document 6] DE102011109440A1 [Patent Document 7] DE102015203889A1 [Patent Document 8] DE102014117845A1 [Patent Document 9] DE102017117594A1 [Patent Document 10] DE102020000292A1 Summary of the Invention [Problem to be solved by the invention]

[0010] SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for calibrating vehicle headlights that overcomes the above-mentioned drawbacks. [Means for solving the problem]

[0011] According to the invention, this problem is solved by a method with the features of claim 1, and here in particular with the features described in the characterizing part of claim 1. Advantageous configurations and developments emerge from the claims dependent on claim 1.

[0012] The essence of the method according to the present invention is that in a first step, at least three images are taken, namely, a first image including the calibration pattern, a second image without the calibration pattern, and a third image including the inverted calibration pattern. In a second step, the second image is subtracted from the first and third images to clean up the images from the existing scene. The calibration pattern includes circles, whose positions in the image are known, and whose centers are located on the horizon. A linear approximation is used to calculate the centers of each circle and determine whether the projection of the calibration pattern is in the correct plane for performing the calibration. As a result of the image subtraction, only the calibration pattern is visible in the image at any one time. This significantly simplifies the subsequent evaluation algorithm and makes it more robust. Thus, a system for automatically and accurately calibrating the adjustment of a high-resolution headlight system in a vehicle is provided.

[0013] The method for calibrating vehicle headlights determines misalignment of a predetermined light distribution by evaluating images from at least one vehicle camera, which can then be compensated for by, for example, an installed stepping motor and by redefining the zero position.

[0014] During calibration, the method can detect structures in the area ahead of the vehicle and, in particular, output whether there is a wall in front of the vehicle, as well as determine the orientation of the wall. All this information can be used to adapt the projection from the headlights to the existing area ahead.

[0015] Preferably, in a third step, the first and third images can be compared pixel by pixel to identify brightness differences between pixels at the same location.

[0016] Here, according to an advantageous configuration, areas with very small brightness differences can be represented as gray areas. For example, areas outside the calibration pattern are marked as gray areas, and since these areas are also present in the camera image, i.e., in the image, they do not need to be further evaluated. This has the advantage that the background scene is represented by gray values ​​that are clearly separated from the headlight projection area. Furthermore, the fusion of the two inverse images can eliminate the influence of optical crosstalk of active pixels due to imperfect optical channels. This can improve, for example, the localization of subsequent circles.

[0017] The calibration pattern includes a circle, whose position in the image is known. High-resolution systems allow for the use of very different projections for calibration. Thus, for example, circles can be used, which offer several advantages. For example, circles have a constant center point, independent of the focus of the illumination system on the projection plane. Circles are also robust to distortions.

[0018] The center points of the circles are located on a horizontal line. Advantageously, the position of the circles in the headlight illumination area is known. This can be done, for example, by using a known spacing between the parallel lines and a maximum allowable distance or a known distance between the center points.

[0019] A linear approximation is used to calculate the center point of each circle, and it is determined whether the projection of the calibration pattern is on an appropriate plane for performing the calibration. For example, the required maximum rotation and the spacing of parallel lines can ensure that the projection is on a plane. Only in such cases is the calibration advantageously continued.

[0020] According to a very advantageous development of this idea, vehicle camera can be pre-calibrated, which is advantageous for the implementation of the above-mentioned linear approximation.Pre-calibrated vehicle camera is also advantageous for determining the position and rotation of projection plane, i.e., for determining the pose.

[0021] According to an advantageous configuration, the orientation of the projection plane can be determined by rotating the calibration pattern horizontally and vertically and shifting the center point of the circle in the camera image to calculate the orientation of the projection plane. This is done, in particular, by assuming a flat projection plane. In a final step, the orientation of the headlight can be determined via the orientation of the plane, the position of the center point in the camera image, and the known angular position of the associated headlight pixel, and the associated positional deviation can be derived. The known angular position can be the vertical angle relative to the center point and / or the horizontal angle relative to the center point of the circle.

[0022] Similarly or alternatively, the headlights can be pre-calibrated to use the circle as a feature point for triangulation of 3D coordinates. Preferably, an additional 2x3 image is taken once for each of the two shifted calibration patterns. In this case, a corresponding camera-headlight pair is used as the structured light system.

[0023] Similarly, it is conceivable to model the projected circle as the side of a cone when expanding it in space. Here too, the position of the ring in three-dimensional space can be accurately determined, and since the mounting position of the headlight is known and the light beam can thus be reconstructed, the adjustment of the headlight can be directly calculated backward.

[0024] In another advantageous configuration, the pose can be determined by the known angular position of the calibrated vehicle camera, and flat surfaces and discontinuous scenes can be detected to adapt the projection and / or staged animation to the projection surface. Advantageously, this allows for accurate measurement of the projection surface in front of the vehicle, whereby not only flat surfaces such as walls or roads but also more complex and unstable scenes can be detected.

[0025] Further advantages of the method according to the invention emerge from the remaining dependent claims and on the basis of the exemplary embodiments that are described in more detail below with reference to the drawings. [Brief explanation of the drawings]

[0026] [Figure 1] 1 shows one possible embodiment of a calibration pattern. [Figure 2] 10 illustrates another possible embodiment of the calibration pattern. [Figure 3] A possible progression of the method is shown. DETAILED DESCRIPTION OF THE INVENTION

[0027] 1 shows a possible embodiment of a calibration pattern 4. The calibration pattern 4 comprises individual circles 9 each having a center point 10. The surfaces represent, for example, the surfaces of headlights 13.

[0028] Another variant of the method 1 can be seen in Figure 2. The difference with respect to Figure 1 is the different arrangement of the circles 9. Furthermore, horizontal lines 11 running along the center points 10 can be seen. In this type of calibration pattern 4, vertical lines 11' can also be drawn for evaluation, which also connect two center points 10 together.

[0029] FIG. 3 shows a possible sequence of the method 1. In a first step 2, three separate images are generated: a first image 3 is captured, including the calibration pattern 4; a second image 5 is captured, without the calibration pattern 4; and a third image 6 is captured, including the inverted calibration pattern 4'. In a second step 7, the second image 5 is subtracted from the first image 3 and the third image 6. This allows the images 3 and 6 to be cleaned from the separate scenes present. For clarity, the second step 7 is shown duplicated. In a third step 8, the two images 3 and 6 are compared pixel by pixel to find brightness differences between pixels at the same location. Small brightness differences are represented as gray areas, which correspond to the area outside the headlights 13 and are represented by thin dashed lines.

[0030] Therefore, the method realizes simple and automatic headlight calibration without requiring additional hardware components. Advantageously, high accuracy can be achieved. Furthermore, the method is robust to environmental conditions and can be performed autonomously, thus without the involvement of the driver or other operators. This entails safety benefits as well as comfort benefits. Similarly, the projection space profile calculated during operation can be used to adapt the projection of the vehicle, for example, the animation for the start-up scene, to the existing projection surface.

Claims

1. 1. A method (1) for calibrating a vehicle headlight, wherein a misalignment of a predetermined light distribution is determined by evaluating images from at least one vehicle camera, the method comprising: In a first step (2), at least three images are taken, namely a first image (3) including the calibration pattern (4), a second image (5) not including the calibration pattern (4), and a third image (6) including an inverted calibration pattern (4'), and in a second step (7), the second image (5) is subtracted from the first and third images (3, 6) to clean up the images (3, 6) from the existing scene; The method (1), characterized in that the calibration pattern (4) includes circles (9), the positions of the circles (9) in the image are known, the center points (10) of the circles (9) are located on a horizontal line (11), the center points (10) of each circle are calculated using linear approximation, and it is determined whether the projection of the calibration pattern (4) is made in an appropriate plane suitable for performing the calibration.

2. 2. The method (1) according to claim 1, characterized in that in a third step (8) the first image (3) and the third image (6) are compared pixel by pixel in order to identify brightness differences between pixels at the same location.

3. 3. Method (1) according to claim 2, characterized in that areas with very small luminance differences are represented as grey areas.

4. Method (1) according to any one of claims 1 to 3, characterized in that the vehicle camera is pre-calibrated.

5. 5. The method (1) according to claim 1 or 4, characterized in that the orientation of the projection plane is determined by rotating the calibration pattern (4) horizontally and vertically and shifting the center point (10) of the circle in the camera image (12) to calculate the orientation of the projection plane.

6. The method (1) according to claims 4 and 5, characterized in that the pose is determined by a known angular position of the calibrated vehicle camera, and that detecting flat surfaces and discontinuous scenes allows for adapting the projection and / or staged animation to the projection surface.

Images