DE102011109440A1 - Method for adjusting and / or calibrating at least one headlight of a vehicle - Google Patents

Abstract

The invention relates to a method for calibrating and / or adjusting at least one headlight and / or a light unit (4) of a vehicle (1), wherein by means of a detection unit, in particular an image detection unit (2), images (X) of a front of the vehicle (1) located and detected by means of the at least one headlight or the light unit (4) illuminated vehicle environment. According to the invention, the images (X) are read out pixel by pixel and while maintaining row and / or column information, a matrix structure is determined in which the horizontal and / or vertical of a light-dark boundary (HDG) and a deviation (ΔA, ΔL) of the light -Darkel limit (HDG) are determined from a predetermined setpoint (ASoll, LSoll), the headlights or the light unit (4) being calibrated or adjusted on the basis of the deviation (ΔA, ΔL) determined.

Description

The invention relates to a method for adjusting and / or calibrating at least one headlight and / or a light unit of a vehicle, wherein captured by means of a detection unit, in particular an image capture unit images of a vehicle located in front of the vehicle and illuminated by means of at least one headlamp or a light unit vehicle environment become.

Various methods for calibrating and adjusting headlamps of a vehicle are known from the prior art.

From the DE 196 02 005 A1 For example, an apparatus and a method are known for automatically correcting an alignment of at least one headlight of a motor vehicle when the position of the vehicle changes. The apparatus includes an illuminance sensor, an actuator for stiffening the tilt of the headlamp relative to the vehicle, and processing means for establishing a control signal for the actuator in response to the signals generated by the sensor. The device further comprises a radiator that forms a light spot on the floor in front of the vehicle, wherein the sensor and the radiator are arranged on the vehicle such that the sensor observes a shift of the light spot in relation to its detection field when the vehicle changes position. The processing means generates the control signal in dependence on the shift observed by the sensor. Depending on the position of the detected light spot, the orientation of the at least one headlight is adjusted.

Furthermore, from the DE 43 41 409 A1 a device and a method for controlling a headlight range of headlights of motor vehicles known. The device comprises a transmitter emitting electromagnetic radiation which generates a low-expansion radiation spot on a roadway at an actual distance from the motor vehicle. If the headlamp range of the headlamp is correctly set, the area is at a correct distance from the vehicle. Furthermore, a sensor which detects radiation reflected from the roadway and an evaluation device connected thereto are present. The sensor has a plurality of sensitive to the radiation emitted by the transmitter elements and an imaging optics, by which the incident radiation is directed depending on the angle of incidence on one of the elements. Based on the angle of incidence, a deviation of the actual distance from the correct distance is determined. By the evaluation device is determined which of the elements incident radiation hits, and depending on the headlamp associated adjustment is controlled to set the actual distance and thus the headlight range to a predetermined value, the transmitter and thus the actual distance mitverstellt ,

From the prior art, as in the DE 10 2007 049 619 A1 described a method for automatic adjustment of a light unit of a vehicle and a lighting system for a vehicle. In the method, at least one, a specific light function characterizing light distribution of the light unit is generated by means of at least one activated base light source. In this case, at least one optical adjustment mark is generated in the visible spectral range invisible to the human eye. The adjustment mark can be detected in an adjustment mode of the light unit by means of a detector unit. Depending on detecting the adjustment mark and comparing the adjustment mark with a reference mark, the light unit is adjusted.

In the DE 199 02 015 A1 An arrangement for adapting the lighting system to a motor vehicle is described. The arrangement comprises a device for detecting the traffic situation in front of the motor vehicle and an adjusting device for the lighting system. The traffic situation in front of the motor vehicle is detected by means of an image capture device. The captured images are fed to an image evaluation device, which controls the adjustment device for the illumination system according to the respective requirements on the basis of the acquired images. Quantities detectable by the image capture device are approaching traffic, headlamp contamination, ambient brightness, tunnels, rain, fog, snowfall, wet or dry road, type of road, head and tail traffic, vehicle pitch, knolls and valleys, headlight misadjustments, and curves and curves.

From the DE 41 22 531 A1 a device for controlling the headlight range of headlights of motor vehicles is known. The device has arranged on the front of the motor vehicle optoelectronic sensors, an evaluation and the headlamps associated adjustment. By means of the sensors, the illumination of the roadway in front of the motor vehicle is registered in the form of the light reflected by the roadway of the headlights. The sensors have photosensitive elements, by which the position of the illumination limit of the illumination of the roadway can be detected. The sensors conduct the evaluation device signals about the actual illumination of the road to, wherein in the evaluation device a target illumination is stored in the form of data. In the evaluation device, the signals of the sensors are compared via the actual illumination with the stored desired data and driven depending on the detected deviation, the adjustment by which the headlights are adjusted until the actual illumination with the least possible deviation of the target illumination equivalent.

Such a method discloses the US 5 633 710 A1 , wherein by means of an actuator, an alignment of a headlight is set in response to a control signal. In the headlight, an infrared light source is integrated, which generates an infrared light beam and radiates toward a ground surface, wherein by means of an optical sensor system from the ground surface reflected infrared light radiation is detected. The optical sensor system comprises a plurality of sensors, by means of which a plurality of intensities of the reflected infrared light radiation are detected. With the sensor system, a control unit is coupled, by means of which the control signal is formed. The control signal is provided to adjust the orientation of the headlamp so that the intensities of the reflected infrared light radiation correspond to a predetermined state. This condition coincides with a desired orientation of the headlamp to realize automatic alignment of the headlamp.

The DE 10 2010 048 689 A1 whose priority is claimed here, describes a method for calibrating and adjusting a headlamp of a vehicle, wherein by means of a detection unit, a defined light pattern generated in the vicinity of the vehicle is detected and calibrated and adjusted based on the position of the light pattern relative to the vehicle headlights becomes. By means of the detection unit, a first reference image of a low beam distribution and a second reference image of a high beam distribution are detected, wherein the first reference image is detected immediately before switching and the second reference image immediately after the switching between the low beam distribution and the high beam distribution. A common correlation image is generated from the reference images, wherein a global position of at least one light-dark boundary of the low-beam light distribution forming the light pattern is determined on the basis of contrast ratios between the low beam distribution and the high beam distribution. The determined global position is compared with a desired position and, if a deviation is determined, the headlight is calibrated and adjusted.

From the DE 10 2009 047 932 A1 a headlamp for a motor vehicle is known in which for adjusting the headlamp at least one light module emits light in a predetermined light axis and a device for basic adjustment of the light module is designed as an electromotive device, the position of a asymmetry in a cut-off by means of an image acquisition camera is determined and the basic adjustment of the light module takes place in dependence on the specific position by means of the electromotive device.

From the DE 10 2005 040 980 A1 It is known to detect by means of a digital camera or a sensor matrix, the light-dark boundary of a headlamp setting on a projection screen to capture. The camera is hereby separated from the vehicle and serves as the basic setting of the bright dark limit when the vehicle is stationary. For this purpose, the image recorded by the projection screen is evaluated with a computer to derive control commands for the headlamp setting and to achieve a basic adjustment of the headlights.

The invention has for its object to provide a comparison with the prior art improved method for adjusting and / or calibrating at least one headlight and / or a light unit of a vehicle.

The object is achieved by a method having the features specified in claim 1.

Advantageous embodiments of the invention are the subject of the dependent claims.

According to the invention, a method for adjusting at least one headlight and / or a light unit (US Pat. 4 ) of a vehicle ( 1 ), wherein by means of a detection unit, in particular an image acquisition unit ( 2 ) one or more images (X) of one in front of the vehicle ( 1 ) and by means of the at least one headlight or the light unit ( 4 ) illuminated vehicle environment are detected, and in the one or more images (X) while maintaining a row and / or column information, the position of a cut-off (HDG) determined and compared with a desired position (A _Soll , L _Soll ), being based on a determined deviation (.DELTA.A, .DELTA.L) between position and desired position, the adjustment of the headlight or the light unit ( 4 ) is changed,
and by appropriate measures the contrast ratio of the bright-dark border (HDG) is changed.

By influencing the contrast ratio of the bright-dark border, this can be better resolved by the image capture means. This greatly improves the accuracy and reliability of the process.

The contrast ratio can be changed in particular by overlaying several images, by switching between different light distributions or by image processing by means of filter operators (FO). Filter operators can in particular increase the contrast ratio or focus on a few image lines. By switching the light distributions, the detection of a bright-dark border is possible even in difficult lighting conditions with low contrasts.

In a method for calibrating and / or adjusting at least one headlight and / or a light unit of a vehicle, images of a vehicle environment located in front of the vehicle and illuminated by means of the at least one headlight or the light unit are detected by means of a detection unit.

According to the invention, the images are read out pixel by pixel, while maintaining a row and / or column information, a matrix structure is determined in which vertical and / or horizontal of a cut-off and a deviation of the cut-off and a predetermined setpoint are determined , wherein calibrated or adjusted based on the determined deviation of the headlight or the light unit.

By determining the verticals and horizontals of the light-dark boundary in the matrix structure of a pixel acquisition unit, in particular an image acquisition unit complex sensors and / or signal generators are avoided and thus a determination of the cut-off line without generating electromagnetic radiation during operation of the vehicle possible. In particular, no changes to the headlight or the light unit are required. Rather, a detection unit already present in the vehicle, in particular an image capture unit, such as a camera, can be used. Thus, the method with little effort and low cost can be realized.

A possible embodiment provides that the images are read out pixel by pixel and stored while maintaining the row and / or column information, the light-dark boundary in the stored pixel image and a deviation of the light-dark boundary in the stored pixel image of a predetermined Setpoint to be determined, being calibrated or adjusted based on the determined deviation of the headlight or the light unit.

Also, the determination of the cut-off line based on stored pixel images allows a simple process with low cost and little effort.

A further embodiment of a method for calibrating and / or adjusting the headlight and / or the light unit provides that in a first method step, the captured images are read out pixel by pixel and stored while maintaining a line information, in a second step, the stored images to total Further processed result images, where light-dark transitions are amplified in the stored images by means of at least one filter operator. Furthermore, in a subsequent third method step, a position of a light-dark boundary of a light distribution generated by the headlamp in a line structure of the overall result image is determined, based on at least one determined image line in which the cut-off line is located, a distance the light-dark boundary of the vehicle and / or a position of the cut-off in the vehicle environment is determined. Furthermore, in a fourth method step, a deviation of the distance and / or the position of the cut-off line from a predetermined desired value is determined in a calibration.

By means of the stepwise method, it is possible in a particularly advantageous manner, without the generation of a specific electromagnetic radiation, in particular without generating a specific light pattern, directly from the light distribution generated by the headlight, the distance of the light-dark boundary to the vehicle and the position of Hell Darkness limit in the vehicle environment to determine and easily determine the calibration of the deviation of the position and / or the distance from the target value. It is particularly advantageous that no additional sensors are required, but already Sensors installed in the vehicle, in particular cameras, such as a camera of a night vision assistance system or a multi-function camera, can be used. Furthermore, results from the pixel and line by line processing and storage of recorded images and the possibility of resulting pixel and line by line evaluation of the overall result images in which the light-dark transitions are amplified by means of the filter operator, a particularly accurate measurement and determination the cut-off line from the overall result images.

In a development of the method according to the invention, the light-dark transitions are focused by means of the filter operator on a maximum of two image lines, so that a particularly high contrast is achieved in the overall result images and consequently the light-dark boundary can be determined very accurately.

According to one embodiment of the method according to the invention, during the adjustment of the headlight and / or the light unit, parameters of at least one servomotor of a headlight range control are adjusted such that the deviation of the distance and / or the position of the light-dark boundary from the given desired value is automatically eliminated. Thus, an automatic adjustment of the headlight or the light unit during the driving operation of the vehicle is possible and always optimal illumination of the vehicle environment feasible.

In order to achieve optimum illumination in front of the vehicle at each light distribution which can be generated by the headlight and / or the light unit, in an expedient embodiment the method steps are carried out cyclically for different settings of the headlight or the light unit and the associated different light distributions.

Particularly preferably, the light distribution is lowered in such a way before the method steps are carried out that a minimum illumination range of the headlight and / or the light unit is set. As a result, the accuracy of detection of the distance and the position of the cut-off line is further increased since a visible projection area of the light distribution is maximized in the image captured by the detection unit. Due to a distance minimization between the detection unit and the light-dark boundary of the light distribution of the headlight or the light unit, both a resolution and a dynamic range within the image are substantially improved. Furthermore, due to the improved resolution, it is possible that the determination of the cut-off line does not have to take place in a subpixel area of the recorded image, which would lead to a high degree of accuracy with increased complexity. Furthermore, a driver of the vehicle is actively informed by lowering the light distribution that a headlight or light unit setting is re-measured. Subsequently, it is preferably output by means of an optical display unit, which is for example part of a so-called instrument cluster of the vehicle, that the current headlight or light unit setting has been calibrated and adjusted and / or corresponds to legal regulations. As a further advantage, it can be seen from the outside whether the vehicle is equipped with a corresponding device for carrying out the method for calibration and / or adjustment.

In an expedient refinement, a barycentric edge operator and / or a Sobel operator are or are used as the filter operator, which deliver particularly accurate results in the determination of gradients within the overall result image and the light-dark boundary determined from the gradients.

In order to enable a statistically particularly robust and reliable determination of the cut-off line, a plurality of filter operators is used to amplify the light-dark transitions in the stored images in the second method step.

In this case, a first result image is preferably determined by means of the filter operators, and then the respective first result image is converted into a second result image by means of a non-linear transfer function, whereby pixel contents below a predetermined brightness threshold value are suppressed disproportionately within the respective first result image and pixel contents above the predetermined brightness threshold value be reinforced disproportionately, the second result images are assembled to the overall result image. In particular, up to 16 filter operators are used, with a particularly robust and reliable determination being achieved.

Furthermore, the filter operators are preferably determined prior to implementation in the method in an optimization method or in a test method, wherein in the optimization method and in the test method for different real lighting scenarios, the respective associated filter operator is determined. As a result, it is always possible to select from the available filter operators the one which maps the light-dark boundary belonging to the respective illumination scenario with the highest reliability in the overall result image and thus enables the most accurate results.

In order to further increase the accuracy of the determination of the cut-off line, the acquired images are normalized and / or folded in a development of the method according to the invention in a mean-variance normalization. The mean-variance normalization enables on the one hand a precise evaluation of image areas with a homogeneous brightness distribution and on the other hand an elimination of occurring disturbances, in particular between the several second result images, which are combined to the overall result image.

A further embodiment of the method for adjusting at least one light unit and / or one headlamp provides that at least one light distribution of the light unit characterizing a specific light function is generated by means of at least one activated light source. In this case, at least one image of the light distribution is detected by means of at least one image acquisition unit of the vehicle and an evaluation of the at least one acquired image in an image evaluation unit determines a horizontal and / or vertical position of at least one predetermined part of the light distribution and with a predetermined horizontal and / or vertical Target position compared, wherein in a deviation of the determined position of the predetermined target position an indication message to a driver is generated and / or means for horizontal and / or vertical adjustment of the headlamp and / or the light unit are controlled.

The method enables a simple and exact check of an adjustment of the light unit and / or the headlight of the vehicle during operation of the vehicle, wherein this check is preferably carried out automatically. If a misalignment of the light unit or the headlight is detected, this is corrected either by an automatic adjustment of the light unit or the headlight or it can, for example, a warning message are generated, which prompts for a workshop visit. In this way, a correct adjustment of the light unit and thus an optimal illumination of an environment of the vehicle are ensured without dazzling other road users. This increases road safety for both the vehicle and other road users. The check can be carried out regularly by means of the method so that an immediate adjustment is made possible in the event of a malposition of the light unit and / or of the headlamp. In this way, a proper function of the light unit and / or the headlamp is ensured, d. H. a maximum possible brightness for a driver with a minimum possible glare other road users.

In particular, a camera arranged on the vehicle can be used as the image capture unit, for example a so-called multi-purpose camera, near-range infrared camera and / or far-range infrared camera, in particular for night vision functions, but also also a radar and / or lidar system. Advantageously, the vehicle has a plurality of such image acquisition units, which are used for the method. For this purpose, in particular image acquisition units can be used, which are already installed on the vehicle for other purposes, in particular an image acquisition unit of a light assistance system of the vehicle to detect objects in an environment of the vehicle, in particular objects in front of the vehicle, for example oncoming or preceding other vehicles. In order to avoid dazzling of these other objects, it is then switched to a different light distribution and / or the respective light distribution is pivoted away from the respective detected object. Also a so-called danger light function, d. H. a targeted illumination of an object with which there is a danger of collision is made possible by means of the light assistance system and the image acquisition unit. Furthermore, the image acquisition unit to be used in the method can also be provided on the vehicle, for example, for environmental detection as a part of a parking assistance device and / or a traffic monitoring and accident prevention device.

Advantageously, no additional alignment mark is produced in the light image of the active light source by means of the at least one active light source, but the already existing light distribution of the active light source is used for the adjustment.

In an easy to implement embodiment of the method, which is suitable for example for adjusting the light unit or a plurality of light units of the vehicle in a workshop, the vehicle is in front of a flat calibration z. B. a wall placed, preferably at a distance from five to ten meters to the wall and possibly such that the wall is arranged substantially perpendicular to a direction of travel of the vehicle.

The calibration screen or the wall preferably has markings, so-called calibration targets or calibration marks, via which the distance of the vehicle to the wall and a possible misalignment of the vehicle to the wall can be determined by means of the image acquisition unit, so that the vehicle is at an exact distance and a exact angle to the wall can be positioned and so that the exact location of the calibration targets in relation to the camera and thus to the vehicle can be determined.

By means of the light units or by means of their light sources, the light distribution is generated and thus the wall is illuminated. In particular, a light-dark boundary of the light distribution on the wall is very rich in contrast and easy to recognize and detect by means of the image acquisition unit. By an exact positioning of the vehicle in front of the wall, the light distribution and thereby also the cut-off for a correctly adjusted light unit at a predetermined location on the wall, which is for example marked by a corresponding marker and which of the image processing unit is known. This can be determined by evaluating the image captured by the image acquisition unit. If it is determined by the evaluation of the image that the light unit is not aligned correctly, it can be adjusted horizontally and / or vertically on the basis of the marking on the wall in a simple manner. In the workshop this is possible for example by a manual adjustment.

For example, to eliminate human error and to simplify the adjustment, an automatic adjustment makes sense. For this purpose, corresponding means for horizontal and / or vertical adjustment of the light units are activated and the light units are thereby automatically adjusted. On the one hand, the camera is controlled as a positioning aid for the vehicle in the light adjustment and on the other hand, the automatic evaluation of the light / dark boundary, in which the corresponding deviations are stored.

In a preferred embodiment of the method, before the at least one image of the light distribution is detected, an optical axis of the at least one light unit is vertically pivoted from a starting position by a predetermined value by means of the at least one image acquisition unit of the vehicle. The starting position may be, for example, a normal position of the light unit, in which the light distribution or specific features of the light distribution, such as the cut-off, reflected at a relatively large distance in front of the vehicle and with a relatively low contrast of a road surface.

The optical axis is pivoted so far vertically from this initial position that the light distribution of the light unit is reflected as close to the vehicle, but still in a detection range of the image acquisition unit of the road surface. In this way, the light distribution and in particular their specific features are much more rich in contrast detected by the image acquisition unit and evaluable by the evaluation unit. The vertical pivoting of the optical axis is performed for example by means of an automatic headlamp leveling of the vehicle, which is already installed in vehicles with gas discharge lamps or LED bulbs in the vehicle. This headlamp leveling is usually carried out by means of stepper motors, so that the headlight range can be set very accurately and in this way, the optical axis of the light unit can be pivoted by the predetermined value. After carrying out the procedure, d. H. After checking the position of the light distribution and optionally after an automatic adjustment of the light unit, the optical axis is pivoted back to the normal position vertically, so that the light unit is ready for a driving operation of the vehicle.

Appropriately, before determining the horizontal and / or vertical position of at least one predetermined part of the light distribution is checked whether a relevant detection range of the image acquisition unit is free of obstacles that would distort the determination of the horizontal and / or vertical position of at least a predetermined part of the light distribution. If such obstacles are detected, no position determination of the light distribution is performed. If the light distribution and in particular their specific features are not reflected by the flat road surface, but by obstacles, possibly false positions of the specific features are determined, thereby diagnosing a misalignment of the light unit, which, however, is not real. This could lead to a misalignment of the light unit, which is prevented by detecting the obstacles.

Preferably, the method is performed immediately after an engine start of the vehicle. In this way, the light unit of the vehicle is checked at each departure and either automatically adjusted, if necessary, or the driver at least pointed to a required adjustment. This avoids prolonged driving with a wrongly adjusted light unit, thereby avoiding a non-optimal brightness for the driver and dazzling other road users. The method is preferably carried out immediately after the engine start and a self-referencing of the light unit taking place immediately thereafter. Such self-referencing is performed on vehicles having automatic headlamp leveling to adjust a headlamp range corresponding to a vehicle height and inclination.

In particular, when during the implementation of the method, the optical axis of the light unit is pivoted from the starting position, for example, from the normal position, vertically downwards, the method is advantageously carried out only when the vehicle is stationary, since in this case the environment of the vehicle only is lit very limited. When the vehicle is moving, this would lead to a risk to vehicle occupants of the vehicle and other road users due to lack of visibility of the driver.

Alternatively, the method is performed only when the vehicle moves in the dark in a straight line and uniformly on a dry road. This condition can easily be determined with existing sensors (inertial sensor and rain-light sensor). This is possible, in particular, when the optical axis of the light unit is not pivoted vertically by the predetermined value during the method, since illumination of the surroundings of the vehicle with the light unit is then not or only insignificantly restricted during the execution of the method. An advantage of this solution is that a driving operation of the vehicle is not hindered. For example, the driver need not wait until the process is complete and the optical axis of the light unit is returned to normal before it can move the vehicle. This makes it possible to carry out the process regularly and in this way to check the exact positioning of the light distribution at regular and short intervals and adjust automatically if necessary. The rectilinear uniform movement in darkness allows a good detection of the light distribution by means of the image acquisition unit and their good and accurate evaluation in a simple manner.

As light distribution, preferably a low-beam light distribution, a vertical beam of the optical axis in the direction of the vehicle shifted Abblendlichtverteilung or shifted in this way other light distributions, a highway distribution, a Teilfernlichtverteilung, a high beam distribution, a motorway light distribution, a hazard distribution and / or a crosshair distribution generated. These light distributions each characterize specific light functions of the light unit. The light distributions can be generated for example by means of swiveling light sources and / or light units, by means of diaphragms and / or rollers in the light units and / or by means of a plurality of light sources, which are then preferably designed as light-emitting diodes (LED) and expediently in a so-called LED Array are arranged, d. H. For example, a plurality of light-emitting diodes are arranged side by side and a plurality of rows of juxtaposed light-emitting diodes with each other, wherein preferably each of the light-emitting diodes is individually switchable.

Furthermore, a multiplicity of further light distributions can be generated by means of such LED arrays, by means of which accurate horizontal and / or vertical position determination of the respective light distribution and thereby exact adjustment of the light unit is possible. In particular, the crosshair light distribution can be generated in a simple manner by means of such LED arrays. With this reticle light distribution, the position of the light distribution can be determined in a particularly simple and exact manner, and thus the light unit can be precisely adjusted in a particularly simple manner. The light distributions can be aligned horizontally and / or vertically by using means for horizontal and / or vertical adjustment of the light units, the means being, for example, electric motors and a corresponding mechanism for automatic headlamp leveling and / or horizontal and / or vertical pivoting of the light units ,

Advantageously, a first image of the motorway light distribution is detected, switched to the highway distribution and a second image and switched back to the motorway light distribution and a third image captured, wherein by means of the image evaluation unit an average image of the first and third image is formed and a difference image of the mean and of the second image is determined. The formation of the averaged image reduces changes in scenery during image capture. After formation of the difference, ie the difference image of the motorway light distribution and the country road light distribution, only a bright spot remain in a center of the difference image of the motorway light distribution and the country road light distribution and a distant bright area, which consists of a bright additional share of Highway light distribution over the highway distribution result, since the difference in the same bright components disappear.

In addition, inhomogeneities, in particular markings traveled in a measuring interval, are retained, but these are easily recognizable by the image evaluation unit by means of corresponding image processing algorithms. in particular, the desired position of a center of gravity of the bright spot is known and can be compared with the currently determined position in the difference image of the motorway light distribution and the country road light distribution. In addition, averaging over a plurality of differential images of the highway light distribution and the highway distribution thus obtained can be made to increase accuracy and remove inhomogeneities. The short-term shutdown of the motorway light distribution is barely perceptible, nevertheless, the driver is preferably before a corresponding notice, for example, displayed in an instrument cluster of the vehicle in order not to irritate this by the deactivation.

If the determined vertical position of the center of gravity of the bright spot in the differential image of the motorway light distribution and the country road light distribution do not coincide with the corresponding predetermined target position, then here also the information message to the driver is generated so that he can visit a workshop and can adjust the light unit , or the means for horizontal and / or vertical adjustment of the light units are controlled and the light units are automatically adjusted in this way.

In an advantageous embodiment of the method, a plurality of images is acquired and an average value is formed from the captured images or from at least one image region of the captured images. As a result of this averaging of a plurality of acquired images, inhomogeneities are iteratively removed from the mean value image, while constant portions, in particular the cut-off line, emerge more clearly. In particular, the position of a horizontal portion of the cut-off line can then be evaluated and determined by means of conventional image analysis means, such as a differentiation of an intensity distribution in the average image, by the image evaluation unit.

In an advantageous variant of the method is thereby switched from the split-beam distribution in the high-beam distribution or vice versa, with a first image before switching and a second image is detected after switching. Also, the horizontal alignment of the light unit is made possible by a horizontal position determination of the Teilfernlichtverteilung while driving the vehicle and possibly allows their adjustment. If the requirements are met, d. That is, the vehicle is in the vehicle state "straightforward uniform motion in darkness on a dry road" and further, either the sub-beam distribution is switched to the high beam distribution or vice versa, the image capturing unit acquires a first image before switching and a second image after switching , A desired position of vertical shadow edges is predetermined. Thereafter, a difference image from the two images of the Teilfernlichtverteilung and the high beam distribution is formed by means of the image evaluation unit, whereby the same proportions of the two images disappear and are no longer displayed in the difference image of the Teilfernlichtverteilung and the high beam distribution. As a result, the vertical shadow edges emerge strongly, so that their position can be determined by image evaluation of the difference image of the partial high-beam distribution and the high-beam distribution in the image evaluation unit. Furthermore, the position of a horizontal shadow edge can be determined in this way.

If the determined vertical and / or horizontal positions of the cut-off line or the shadow edges do not coincide with corresponding predetermined target positions, then here also the information message is generated to the driver, so that he can visit a workshop and adjust the light units can leave, or it will be driven the means for horizontal and / or vertical adjustment of the light units and the light units are automatically adjusted in this way.

In a workshop, with a flat calibration screen whose flat surface also the illumination range of the headlights and / or the light unit ( 4 ), an adjustment takes place. In this case, calibration marks are used for determining the position of the planar calibration screen relative to the camera or image acquisition unit, and the position of the current light distribution of the at least one lighting unit, light unit or headlight is determined on the flat surface and compared with a desired position.

A further embodiment provides that a defined light pattern generated in the surroundings of the vehicle is detected by means of the detection unit, the headlight and / or the light unit being calibrated and / or adjusted based on the position of the light pattern relative to the vehicle.

In this case, a first reference image of a low beam distribution and a second reference image of a high beam distribution are detected by the detection unit, wherein the first reference image is detected immediately before switching and the second reference image immediately after switching between the low beam distribution and the high beam distribution and generates a common correlation image from the reference images is determined, wherein based on contrast ratios between the low-beam distribution and the high beam distribution in the correlation image, a global position of at least one light pattern forming light-dark boundary of the low beam distribution, wherein the determined global position is compared with a desired position and at a thereby determined Deviation of the headlights and / or the light unit is calibrated and / or adjusted.

The further developed method allows a continuous adjustment and calibration of the headlight and / or the light unit during operation of the vehicle. Thus, an optimal adjustment of the headlamp or the light unit and, consequently, an optimal illumination of the surroundings of the vehicle are always ensured. Due to this optimal illumination, a driver's view of the vehicle while avoiding glare from other road users, in particular road users in oncoming traffic of the vehicle, is maximized. To carry out the method according to the invention, no changes to the headlight or the light unit are required. Also, no special and expensive special sensors for detecting the light distribution is required, since an existing in-vehicle night vision camera and / or a multi-function camera can be used to capture the environment. Thus, the method with little effort and low cost can be realized.

Embodiments of the invention are explained in more detail below with reference to drawings.

Showing:

1 schematically a first reference image of an environment in front of a vehicle with activated low beam distribution,

2 schematically a second reference image of the environment located in front of the vehicle according to 1 with activated high beam distribution,

3 schematically a correlation image generated from the reference images,

4 schematically a first image section of the correlation image according to 3 .

5 schematically a second image section of a correlation image,

6 schematically a from the correlation image according to 3 by means of a low-pass filter determined third image detail,

7 FIG. 2 schematically shows a profile of a coincidence of an angle of inclination of at least one headlight with acquired image information, FIG.

8th FIG. 2 schematically shows a profile of a global position of a cut-off line of the low-beam light distribution as a function of a distance from the headlight, FIG.

9 schematically a profile of the global position of the light-dark boundary of the low-beam light distribution as a function of a tilt angle of the headlamp.

10 a schematic representation of a vehicle,

11 a schematic representation of a Teilfernlichtverteilung on a wall,

12 a schematic representation of a highway distribution,

13 a schematic representation of a motorway light distribution,

14 a schematic representation of a differential light distribution,

15 a schematic representation of a Teilfernlichtverteilung,

16 a schematic representation of a vertical pivoting of an optical axis of a light unit,

17 a schematic representation of a resulting from a pivoting of an optical axis shifted Abblendlichtverteilung,

18 schematically a flow chart of the method according to the invention,

19 FIG. 2 schematically shows a comparison between a distance of a light-dark boundary of a light distribution to a headlight of a vehicle and its imaging in a picture acquired by means of a detection unit, FIG.

20 schematically a comparison of a raw value control of actuators of a headlight range control and a headlight range for four light modules of a headlight pair,

21 schematically a comparison between a measured headlight range and a headlight range control of a light module,

22 schematically a means of the detection unit according to 18 when driving a vehicle at a tilt angle of the headlamp of 4.5% recorded image,

23 schematically a means of the detection unit according to 18 when driving a vehicle with a tilt angle of the headlight of 6.0% recorded image,

24 schematically a processed from a recorded image overall result image and

25 schematically an overall result image of a barycentric edge operator averaged over five result images.

Corresponding parts are provided in all figures with the same reference numerals.

In 1 A first reference image B1 of an environment located in front of a vehicle, not shown, is shown when the low-beam distribution AV is activated. 2 shows a second reference image B2 of the environment in front of the vehicle when the high beam distribution FV and in 3 a correlation image KB generated from the reference images B1, B2 is shown.

The reference images B1, B2 are recorded by means of a not shown, arranged on the vehicle detection unit. The detection unit is a camera and, in particular, part of a driver assistance device of the vehicle, for example a lane assistant, night vision assistant, parking assistant, a distance control system and / or a device for detecting traffic signs.

The low beam distribution AV and the high beam distribution FV shown are generated by means of two non-illustrated headlights of the vehicle. The headlights have one or more lamps, the lamps are incandescent lamps, halogen lamps, gas discharge lamps and / or light emitting diodes.

In order to achieve optimum illumination in front of the vehicle with maximum visibility for the driver while avoiding glare from other road users, it is necessary that the headlights are optimally aligned, d. H. are adjusted.

For this purpose, the headlights during manufacture of the vehicle or during maintenance by means of a setting device, a so-called light box, adjusted. The setting device is mounted in a defined position in front of or on the unloaded vehicle. By comparing a headlight beam with a marker on the setting a deviation is determined, which is minimized by manual adjustment of screws on the headlight housing.

Furthermore, a vertical position sensor for headlamp leveling, which can be operated from the driver's seat, is provided on the vehicle. By means of the position sensor is a light beam of the headlights generated light in particular different loadings of the vehicle and resulting lowering of the tail adjusted.

If the vehicle is equipped with intense light sources, such as headlamps with xenon illuminants, an automatic adjustment of the headlamp range, by means of which additionally higher dynamic vehicle pitch angle changes when accelerating and decelerating the vehicle are compensated.

Furthermore, by means of the headlamp in a manner not shown a Teilfernlicht and a danger marking light can be generated, wherein the optimal function of this in addition to the vertical and a horizontal adjustment or adjustment of the headlight is performed.

Since the adjustment of the headlights during driving of the vehicle, in particular due to vibration and shock changes, an automatic adjustment is performed during operation of the vehicle, so there is always an optimal adjustment of the headlights.

In order to adjust the headlights both horizontally and vertically, a calibration of the same is first required. For this purpose, a defined light pattern on the roadway in front of the vehicle is generated by means of the headlights, which is detected by means of the detection unit. Based on the position of the light pattern relative to the vehicle, the headlights are calibrated.

One possible situation in which a high contrast occurs to measure the horizontal adjustment of the headlights is an adaptation of an adaptive cornering light of the vehicle. Since the headlight control unit is aware in this situation, on which squint angle both headlights are set, this situation can be exploited to determine the basic position of the headlight squint angle.

In particular, a change between the low beam distribution AV and the high beam distribution FV is suitable for measuring the vertical adjustment, wherein the change between the low beam distribution AV and the high beam distribution FV can also be used to determine the horizontal adjustment of the headlights. For the robust measurement of jumps in a local or temporal light distribution, it is particularly advantageous to select a situation with the greatest possible contrast. This contrast is given in the change between the low beam distribution AV and the high beam distribution FV.

Furthermore, it is particularly advantageous that due to the use of the change between the low beam distribution AV and the high beam distribution FV no particularly excellent calibration pattern or light pattern are additionally generated, which could distract the driver of the vehicle from his driving task.

To calibrate the headlights, the first reference image B1 of the surroundings of the vehicle with the low-beam distribution AV and the second reference image B2 of the surroundings of the vehicle with the high-beam distribution FV are detected by means of the detection unit. The detection takes place in such a way that the first reference image B1 is detected immediately before switching and the second reference image B2 immediately after the switching between the low beam distribution AV and the high beam distribution FV.

Switching between the low-beam distribution AV and the high-beam distribution FV takes place automatically by means of a so-called light assistance device, wherein by means of a camera, in particular the detection unit for detecting the reference images B1, B2, the scene is detected in front of the vehicle and then evaluated by means of an evaluation. With a possible glare other road users takes place automatically, d. H. without intervention of the driver, the adjustment of the light distribution of the headlights. By means of such light assist devices, high-frequency changes in the exterior lighting are brought about, which are particularly suitable for use in the adjustment of the headlights according to the inventive method and its embodiments described below, since a particularly robust measurement of the headlight adjustment is made possible by a multiplicity of measured values in conjunction with a low-pass filter is feasible.

Alternatively or additionally, activation and deactivation of the low-beam distribution AV and the high-beam distribution FV is possible manually, in particular via a steering column switch, by the driver of the vehicle.

Both in the automatic adjustment of the light distribution and in the manual operation of the steering column switch, a signal is transmitted to a control unit, not shown, before the actual switching between the low-beam distribution AV and the high beam distribution FV, which drives the detection unit for receiving the first reference image B1. The switchover from the dimming distribution AV to the high-beam distribution FV is delayed, ie only after the detection of the first reference image B1.

Immediately after switching from the low-beam distribution AV to the high-beam distribution FV, the second reference image B2 is detected by means of the detection unit. Alternatively or additionally, the reference images B1, B2 are also detected when switching from the high beam distribution FV to the low beam distribution AV.

In a particularly advantageous embodiment, an environment of the vehicle is detected continuously by means of the detection unit. If an alternation between the low-beam distribution AV and the high-beam distribution FV, in particular from a headlight control unit, is reported, a camera image in the low beam position is marked as first reference image B1 and a camera image in the high beam position is identified as second reference image B2.

From the reference images B1, B2, a common correlation image KB is then generated, in which both the low-beam distribution AV and the high-beam distribution FV are included. Due to the very small time interval between the recording of the images, there are only negligibly small changes in the surroundings of the vehicle, so that the low beam distribution AV and the high beam distribution FV are detected substantially in the same environment.

On the basis of contrast ratios between the low beam distribution AV and the high beam distribution FV, a global position POS _{g of} a light / dark boundary HDG of the low beam distribution AV forming the light pattern is determined in the correlation image KB.

In order to determine the global position POS _{g of} the light-dark boundary HDG, a particularly excellent image section BA1 is marked as a so-called "region of interest" (ROI for short). By comparing the image data from the reference images B1, B2 within the image section BA1, a strong contrast in the image data becomes apparent.

This contrast represents the light-dark border HDG of the headlight cone in the dipped-beam position 4 local position POS _{1 of} the light-dark boundary HDG in the image detail BA1 with a position POS _{B of} the image detail BA1 in the correlation image KB, the global position POS _{g of} the light-dark boundary HDG in the correlation image KB is determined. The determined global position POS _g is then compared with a stored desired position and at a deviation determined therefrom, the headlight is automatically adjusted, for example by means of servomotors.

For this purpose, taking into account the known calibration data of the detection unit used and a so-called ground-plane assumption, i. h, assuming a flat roadway, an in 8th determined distance E of the cut-off line HDG to the headlights determined. By means of this distance E and the knowledge of an installation height of the headlights becomes in 8th calculated closer angle of inclination α of the headlights.

Since a large number of measurement data of the inclination angle α is generated by the continuous evaluation of the image data, an influence of the noise in the reference images B1, B2 and the correlation image KB is minimized by use by means of a low-pass filter or a similar statistical evaluation process. Since the headlight adjustment by mechanical influences, such as vibrations, changed only in the order of several weeks in intensive driving, outweigh the benefits of this statistical compensation calculation against a non-availability of a possible current measurement. By additional evaluation of information of a dynamic headlamp range control, such as a time and an amplitude of the headlamp leveling, thereby caused influences in the measurement of the headlamp setting are compensated in a particularly advantageous manner. One possibility in this case is a correction of the measured headlight setting to the adjustment of the headlight range control. In this case, a measurement can be discarded with active automatic headlight range control, so that only measurements are obtained at leveled vehicle.

By evaluating the information of the headlight range control and additional estimation of a vehicle's own motion, for example based on an evaluation of sensors present in the vehicle, such as a wheel tachometer and / or a steering wheel angle, the ground plane Assumption can be evaluated in a particularly advantageous manner. Since the distance E and a storage of the light-dark boundary HDG are known, it is determined with the help of the vehicle own motion estimation, whether this area is crossed over. By means of a query of the leveling information during the period between measurement and passing over the light-dark boundary HDG is determined whether it was a flat surface. Otherwise, the measurement is preferably discarded.

Furthermore, it is ensured by evaluating information from an automatic proximity control or a high-beam assistant that there is no vehicle between the headlights of the vehicle and the appearance of the light-dark boundary HDG in the image section BA1. Due to the recognition of these dynamic and raised objects, invalid measurements are very easy to discard.

Alternatively or additionally, the headlamp setting is measured with the aid of the vehicles in front. In the case of a flat, largely homogeneous rear front of the preceding vehicle, a projection of the light-dark boundary HDG is detected by contrast detection in the reference images B1, B2 or the correlation image KB1. By means of a distance measurement, for example by means of a sensor system of the adaptive cruise control system, a measurement of the angle of inclination α of the headlights is carried out without temporal recording.

In a particularly advantageous embodiment, a situation in which there is a small distance to the vehicle in front, for example, when waiting at a traffic light, used to determine the inclination angle α, since the small distance disturbances of the measurement are reduced.

4 shows the image detail BA1 in a detailed view, wherein the contrast between the low-beam distribution AV and the high beam distribution FV and the thus very accurately ascertainable bright-dark boundary HDG are illustrated.

5 shows a further image section BA2 in a detailed representation, wherein this image section BA2 shows only a portion of the high beam distribution FV without contrast.

6 shows a filtered image GB determined from the correlation image KB by means of the low-pass filter, wherein noise in the correlation image KB is minimized.

In the following, the calibration and adjustment of the headlights is explained using an example in which the headlamps have light-emitting diodes (= LEDs), in particular a so-called LED array, as the light source.

The headlamps form a freely programmable LED headlamp system, each headlamp has four pixel lines formed from the LED. To realize a strong temporal contrast, the two lowest pixel lines are programmed to a low beam pattern for generating the low beam distribution AV, wherein the lowermost pixel line is alternately turned on and off. A switched-on phase in which the lowermost pixel line is activated or a switched-off phase in which the lowermost pixel line is deactivated has a duration of one second each.

By means of an image processing method, which will be described in more detail below, the light-dark boundary HDG is an edge between the lowest and the second lowest pixel line in the correlation image KB of a vehicle-mounted camera detected and based on the camera calibration the set angle of inclination α of the headlights calculated. The advantage of the implementation shown here is that an explicit definition of the image sections BA1, BA2 is not required, but that the implementation works on the entire correlation image KB.

The first reference image B1 is a recorded camera image when the low-beam light distribution is complete, and the second reference image B2 is a captured camera image when the bottommost pixel line is switched off.

In the calculation of the correlation image KB from the reference images B1, B2, each pixel of the correlation image KB, also referred to as pixel, results from a correlation coefficient of a neighborhood of the size of 10 × 10 pixels of the reference images B1, B2.

Furthermore, a filtered image GB is formed from an initialization from the first reference image B1 and a subsequent erasure of all pixels which have an entry greater than 0.3 at the corresponding location in the correlation image KB. This filtered image GB corresponds to the first reference image B1 without the regions in which the reference images B1, B2 agree to some degree.

Subsequently, all pixels of the filtered image GB that are smaller than the 0.95 quantile of the pixel values of the filtered image GB are set to zero. Thus only the brightly illuminated areas in the filtered image GB are displayed. In other words, in the present case, only the regions in the filtered image GB contain a non-zero entry coincident with the lowest pixel line of the headlight.

To compensate for the movement of the camera by the vehicle's own motion various means are used for correction. For example, an optical flow is taken into account in the calculation of the correlation image KB.

With the aid of the filtered image GB, the present angle of inclination α of the headlights is then determined. For this purpose, a vector is calculated which contains the row sums of the filtered image GB. From the 0.05 and 0.95 quantiles of the vector, upper and lower limits of the lowest pixel row of the LED headlight in the correlation image KB are determined.

By varying the assumed inclination angle α of the headlights, the projection area of the lowest pixel row of the LED headlight in the correlation image KB is calculated by means of the calibration data of the camera. By calculating an intersection area between this projection and the headlight position determined from the vector in the correlation image KB, a correspondence f of the inclination angle α with the image information, a so-called "goodness of fit", is calculated for different inclination angles α.

7 shows an exemplary course of the coincidence f of the inclination angle α with the image information. The maximum of the match f indicates the measured tilt angle α, for which a best possible match f between a headlight model and the evaluated image data results.

In 8th is a curve of the global position POS _{g of} the bright-dark boundary HDG of the low beam distribution AV in the correlation image KB as a function of the distance E from the headlamps shown. By way of example, this illustration shows a relationship between the distance E of the light-dark boundary HDG from the headlights and the resulting image line in the correlation image KB, wherein a measurement accuracy can be estimated from the representation.

If the headlamps are 70 cm high, the HDG cut-off line must be located at a distance E of 70 m from the headlamp. Based on 8th shows that a difference of 6 pixels in the correlation image KB is sufficient to increase the distance E of the light-dark boundary HDG from 70 m to 90 m.

9 shows a course of the global position POS _{g of} the light-dark boundary HDG of the low beam distribution AV in the correlation image KB in dependence on the inclination angle α of the headlights. It can be seen here that the above-described change of the distance E by 20 m or 6 pixels corresponds to a change in the angle of inclination α of the headlights of 0.1. This illustrates that, in a particularly advantageous manner, a temporal smoothing is additionally required, as already described.

10 shows a vehicle 1 with an image capture unit 2 , an image evaluation unit 3 and two light units 4 , The light units 4 each have at least one activatable light source. By means of these light sources are light distributions AL, LL, TL, ABL, ABL 'of the light units 4 which specific light functions of the light units can be generated 4 characterize. Such light distributions AL, LL, TL, ABL, ABL 'are, for example, an in 11 and 15 Partial light distribution shown TL, an in 12 illustrated highway light distribution LL and an in 13 illustrated highway light distribution AL. In the 11 illustrated highway light distribution LL replaced in modern vehicles 1 , which are equipped with a light assistance system, such as an AFS (Advanced Frontlighting System) headlamp, often a in 17 illustrated conventional low beam distribution ABL. Therefore, for the method described below, should the highway distribution LL be unavailable, the conventional low beam distribution ABL may alternatively be used.

The light distributions AL, LL, TL, ABL, ABL 'are, for example, by means of swiveling light sources and / or light units 4 , by means of screens and / or rollers in the light units 4 and / or generated by means of a plurality of light sources, which are then preferably designed as light-emitting diodes (LED) and expediently arranged in a so-called LED array, ie, for example, a plurality of light-emitting diodes next to each other and a plurality of rows next to each other arranged light emitting diodes arranged one below the other, wherein preferably each of the light emitting diodes is individually switchable.

By means of such LED arrays, furthermore, a multiplicity of further light distributions not shown here can be generated, by means of which precise horizontal and / or vertical position determination of the respective light distribution and thereby exact adjustment of the light unit 4 is possible. Such a further light distribution is, for example, a crosshair. The light distributions AL, LL, TL, ABL, ABL are by using means for horizontal and / or vertical adjustment of the light units 4 horizontally and / or vertically alignable, wherein the means, for example, electric motors and a corresponding mechanism for an automatic headlamp leveling and / or for a horizontal and / or vertical pivoting of the light units 4 are.

The image capture unit 2 For example, for the light assistance system of the vehicle 1 used to objects O in an environment of the vehicle 1 , in particular, as in 15 shown, objects O in front of the vehicle 1 to detect, for example, oncoming or preceding other vehicles. In order to avoid dazzling of these other objects O, the system switches to another light distribution AL, LL, TL, ABL, ABL 'and / or the respective light distribution AL, LL, TL, ABL, ABL' from the respective detected object O panned away. Also a so-called danger light function, ie a targeted illumination of an object O, with which there is a risk of collision, by means of the light assistance system and the image detection unit 2 allows.

Such image capture units 2 , which on the vehicle 1 can be used, for example, so-called multi-purpose cameras, short-range infrared cameras, long-range infrared cameras, especially for night vision functions, and also radar and Lidarsysteme. In particular, for such a light assist system is a very precise orientation of the image capture unit 2 or the plurality of image capture units 2 and the light units 4 to each other, so that, for example, in the divided long-distance distribution TL, the detected object O is not dazzled and a non-illuminated area nB of the divided high-light distribution TL, as in 15 represented exactly positioned on the detected object O.

11 shows a first way to adjust the light units 4 of the vehicle 1 by means of one of the light distributions AL, LL, TL, ABL, ABL ', ie in 11 illustrated example by means of the Teilfernlichtverteilung TL. This option is useful, for example, in a workshop. This is the vehicle 1 placed in front of a wall W such that the wall W is perpendicular to a direction of travel of the vehicle 1 is arranged. By means of the light units 4 or by means of their light sources, the partial high-beam distribution TL is generated, and thus the wall W is illuminated. The partial long-distance distribution TL is characterized by the marked non-illuminated area nB, which is in a driving operation of the vehicle 1 on the detected object O in front of the vehicle 1 is aligned to avoid glare of the object O.

The partial high-beam distribution TL has a light-dark boundary HDG, along the non-illuminated area nB a left and a right prominent vertical shadow edge vSKl, vSKr and a striking horizontal shadow edge hSK occur, by means of which the non-illuminated area nB against a lighted surrounding area BU is delimited. Target positions of these shadow edges hSK, vSKl, vSKr are for precisely adjusted light units 4 specified.

By an exact positioning of the vehicle 1 In front of the wall W the partial high-light distribution TL and thereby also the non-illuminated area nB with its vertical shadow edges vSK1, vSKr and its horizontal shadow edge hSK appears in the case of a correctly adjusted light unit 4 at a predetermined position on the wall W. In this illustration, the predetermined position is marked by a cross formed by a horizontal line hL and a vertical line vL, and with a correctly adjusted light unit 4 the horizontal shadow edge hSK of the non-illuminated area nB must lie on the horizontal line hL of the cross and the right vertical shadow edge vSKr of the non-illuminated area nB must lie on a vertical line vL of the cross. In the example shown here is the light unit 4 adjusted correctly. If not, the light unit is 4 by means of the Teilfernlichtverteilung TL and based on the cross on the wall W in a simple manner horizontally and / or vertically adjustable. In the workshop this is possible for example by a manual adjustment.

For example, to eliminate human error and to simplify the adjustment, an automatic adjustment makes sense. This is done by means of the exactly aligned image acquisition unit 2 and the image evaluation unit 3 of the vehicle 1 allows.

In the in 11 In the case illustrated, at least one image of the partial long-distance light distribution TL is obtained by means of the image acquisition unit 2 detected. in the image evaluation unit 3 The positions of the vertical shadow edges vSK1, vSKr and the horizontal shadow edge hSK are then determined and compared with corresponding predetermined target positions. The target positions are given in this example by the lines vL, hL of the cross, but can also be specified by positions of pixels, as in exactly aligned image acquisition unit 2 and light unit 4 the horizontal shadow edge hSK and the vertical shadow edges vSKl, vSKr in each of the image capturing unit 2 captured image must always be positioned in one and the same place. If a horizontal and / or vertical deviation of the positions of the horizontal shadow edge hSK and / or the vertical shadow edges vSKl, vSKr of the non-illuminated area nB in the captured image from the corresponding desired positions is determined, then appropriate means for horizontal and / or vertical adjustment of the light units 4 controlled and the light units 4 thereby automatically adjusted. Since a pixel-accurate determination of the corresponding positions is possible, thereby a very accurate adjustment of the light units 4 allows.

In addition to such adjustment when the vehicle is stationary 1 in a workshop is a regular check of the exact alignment of the light units 4 in a driving operation of the vehicle 1 particularly advantageous to immediately detect a possible incorrect orientation, which can lead to dazzling other road users, for example. Then the light units can 4 be either automatically adjusted immediately or it will at least generate a message to a driver, so that they visit a workshop and the light units 4 can be adjusted.

In the 12 to 15 are various ways to check the exact alignment of the light units 4 of the vehicle 1 when driving on a road FB shown. To make this possible, the image capture unit should 2 a scattered by a road surface light of the light units 4 at a distance of preferably about 50 m to 70 m in front of the vehicle 1 can capture. Upper image areas with oncoming traffic, traffic signs and other such sources of interference are from the image evaluation unit 3 preferably ignored or not with regard to the adjustment of the light units 4 evaluated.

For images of the respective light distribution AL, LL, TL, ABL, ABL 'good with the image capture unit 2 to be able to detect and avoid disturbances, so that an evaluation of the captured images by means of the image evaluation unit 3 and as a result of a vertical and / or horizontal position determination of the respective light distribution AL, LL, TL, ABL, ABL 'in the captured image is possible, the method is performed only when the vehicle 1 in a vehicle state "straight-line uniform motion in darkness" is located. For this purpose, for example, by means of other devices of the vehicle 1 a steering angle or a turning radius, a speed, a condition of the automatic headlight range adjustment, and a brightness.

This can preferably be carried out by means of sensors which are already in the vehicle 1 available. Thus, for example, the steering angle or the curve radius and the speed can be detected by means of sensors of an electronic stability program. A state of the automatic headlamp leveling is for example already to be detected in the prior art when using gas discharge lamps as the light source. The brightness can be detected, for example, by means of a light sensor with which an automatic activation of light functions of the vehicle 1 is possible. Corresponding sensors (control light sensor) can be assumed in modern vehicles, with the corresponding sensor data being accessible via internal bus systems, for example CAN or Flexray.

In 12 an embodiment of the method is shown, by means of which the vertical orientation of the light unit 4 is made possible by a vertical position determination of the highway light distribution LL, so that the headlight range can be determined and, if necessary, is adjustable. If the requirements are met, ie the vehicle 1 is in the vehicle state "straight uniform motion in darkness on dry roads" and further the landing strip light distribution LL is activated, so by means of the image capture unit 2 a first image is taken and in an image memory of the image evaluation unit 3 saved. Thereafter, one or more further images are acquired, for example randomly controlled, and an averaged image is iteratively formed from the acquired images according to the following interactive image processing rule: BS: = (BS + _Bn ) / 2 [1] or: BS: = ((n-1) (n) * BS + B _n / n [2] or, in the case of sufficiently large storage, according to the following rule:

BS here is a respective memory content of the image memory, B _{n is} the currently recorded image and n indicates the number of recorded images. The images BS and B _n represent matrices with corresponding intensity values and the calculation rules relate to all matrix elements.

Other rules of averaging are conceivable. It is important that by this averaging of a plurality of acquired images inhomogeneities are iteratively removed from the averaging image, while constant proportions, in particular the light-dark boundary HDG of the highway distribution LL, stand out more clearly. In particular, the position of a horizontal portion hA of the bright-dark boundary HDG of the highway light distribution LL is then determined by conventional image analysis means such as differentiation of an intensity distribution in the averaged image from the image evaluation unit 3 evaluable and determinable.

In this embodiment of the method, despite the averaging over a plurality of images, the bright-dark boundary HDG and in particular its horizontal portion hA is not clearly visible, this indicates problems on the vehicle 1 towards, for example, a no longer attached light unit 4 or a no longer attached image capture unit 2 , Through such a loose and wobbly light unit 4 no clear highway distribution LL can be generated by a loose and shaky image acquisition unit 2 is the high street light distribution LL or its bright-dark boundary HDG not exactly detectable. Furthermore, there may be further problems with the vehicle 1 to vibrations of the vehicle 1 and as a result lead to a blurred country light distribution LL. This can be determined, for example, based on empirical values and optionally, for example, a request to the vehicle driver for a workshop visit can be generated, and preferably additionally a corresponding indication of the problem in a fault memory of the vehicle 1 registrable.

Since the high street light distribution LL has a characteristic kink K of the light-dark boundary HDG at a predetermined position, besides the vertical orientation, ie the luminous range of the light unit 4 also check their horizontal orientation. If the determined vertical and / or horizontal positions of the light-dark boundary HDG, in particular their horizontal portion hA and their characteristic kinks K do not coincide with corresponding predetermined target positions, then either the information message to the driver is generated, so that this is a workshop can visit and the light units 4 or it can be the means for horizontal and / or vertical adjustment of the light units 4 controlled and the light units 4 are automatically adjusted in this way.

Based on 12 . 13 and 14 a further embodiment of the method is shown, by means of which also the vertical orientation of the light unit 4 is possible, so that the beam range can be determined and, if necessary, is adjustable. When the vehicle 1 in the vehicle state, "straight-line uniform motion in darkness on a dry lane" is located and the highway light distribution AL is activated by the image capture unit 2 captures three images, a first image with activated motorway light distribution AL, as in 12 shown, a second image with deactivated motorway light distribution AL, ie it is first switched to the highway distribution LL, as in 11 and a third image with the motorway light distribution AL activated again, as in FIG 12 shown. From these three captured images is using the image evaluation unit 3 a in 13 shown differential image D of the motorway light distribution AL and the road light distribution LL determined according to the following image processing rule: D: = (a + c) / 2 - b [4] Where a is the first captured image, B is the second captured image, and c is the third captured image.

The formation of the averaging image (a + c) / 2 reduces changes in scenery during image capture. After formation of the difference, ie the difference image D of the motorway light distribution AL and the highway light distribution LL, only a bright spot F remain in an image center of the difference image D of the motorway light distribution AL and the highway light distribution LL and a far away bright area hB, which consists of a bright additional share the highway light distribution AL compared to the highway distribution LL result, since the difference formation equal parts disappear. In addition, inhomogeneities, in particular marks crossed in a measuring interval, are retained, but these are from the image evaluation unit 3 easily recognizable by means of appropriate image processing algorithms. In particular, the desired position of a center of gravity S of the bright spot F is known and can be compared with the currently determined position in the difference image D of the motorway light distribution AL and the country road light distribution LL. In addition, averaging can take place via a plurality of difference images D of the highway light distribution AL and of the highway light distribution LL determined in this way in order to increase accuracy and remove inhomogeneities. The short-term shutdown of the motorway light distribution AL is barely perceptible, nevertheless, preferably the driver before a corresponding note, for example, in an instrument cluster of the vehicle 1 displayed so as not to irritate the deactivation.

If the determined vertical position of the center of gravity S of the bright spot F in the difference image D of the motorway light distribution AL and the country road light distribution LL do not match the corresponding predetermined target position, so here, too, either the information message to the driver is generated, so that he can visit a workshop and the light unit 4 or it can be the means for horizontal and / or vertical adjustment of the light units 4 controlled and the light units 4 are automatically adjusted in this way.

in 15 an embodiment of the method is shown, by means of which using the partial high-beam distribution TL in the driving operation of the vehicle 1 the horizontal orientation of the light units 4 is made possible by a horizontal position determination of the Teilfernlichtverteilung TL and optionally their adjustment is possible. If the requirements are met, ie the vehicle 1 is in the vehicle state "rectilinearly uniform motion in darkness on a dry road" and further is switched either from the Teilfernlichtverteilung TL in the high beam distribution or vice versa, detects the image acquisition unit 2 a first picture before switching and a second picture after switching. The desired position of the vertical shadow edges vSKl, vSKr is predetermined. Thereafter, by means of the image evaluation unit 3 formed a difference image from the two images of the Teilfernlichtverteilung TL and the high beam distribution, with the same proportions of the two images disappear and are no longer displayed in the difference image of the Teilfernlichtverteilung TL and the high beam distribution. As a result, the vertical shadow edges vSK1, vSKr are prominent, as in FIG 15 represented, so that their position by image evaluation of the difference image of the Teilfernlichtverteilung TL and the high beam distribution in the image evaluation unit 3 can be determined. Furthermore, the position of the horizontal shadow edge hSK can be determined in this way.

If the determined vertical and / or horizontal positions of the light-dark boundary HDG or the shadow edges vSKl, vSKr, hSK do not coincide with corresponding predefined setpoint positions, then here as well the notification message is generated to the driver, so that he is a workshop can visit and the light units 4 or it can be the means for horizontal and / or vertical adjustment of the light units 4 controlled and the light units 4 are automatically adjusted in this way.

In this case too, unclear positions of the shadow edges vSK1, vSKr, hSK or the light-dark boundary HDG, because they are not clearly visible, point to the problems already described on the vehicle 1 such as a no longer attached light unit 4 or a no longer attached image capture unit 2 , so that here, for example, if necessary, for example, a request to the driver for a workshop visit generated and preferably in addition a corresponding indication of the problem in a fault memory of the vehicle 1 is registrable.

As in 15 are indicated by the image acquisition unit 2 collected images composed of individual pixels, which are arranged in image lines BPZ and image columns BPS. In order to increase the accuracy of the image evaluation and thereby the position determination of the respective light distributions AL, LL, TL, ABL, ABL 'and / or a processing time and / or a required computing power of the image evaluation unit 3 Therefore, it is useful in a further advantageous embodiment of the method, not all of the image acquisition unit 2 to capture and evaluate captured images, but only in each case predetermined image lines BPZ and / or image columns BPS. This is possible since the shadow edges vSK1, vSKr, hSK or the light-dark boundaries HDG in the light distributions AL, LL, TL, ABL, ABL 'run substantially vertically or horizontally. Thus, for example, in the case of the sub-distance light distribution TL, in which the vertical shadow edges vSK1, vSKr are substantially vertical, an averaging is required only in an expected range of the position of the vertical shadow edges vSK1, vSKr over several image lines BPZ, around a horizontal position of the respective sub-distance light distribution TL to investigate. As the chiaroscuro limit HDG in the highway light distribution LL on a left side of the vehicle 1 , ie, the horizontal portion hA is substantially horizontal, the average value can be formed here over image columns BPS and only these image columns BPS are in the image evaluation unit 3 evaluated.

Another way to get at detected misaligned light units 4 This, at least partially automatically, to adjust exactly, includes involvement of the driver. The driver is requested to perform a light unit referencing. To do this, for example, at a constant speed of 120 km / h on a dark motorway, the driver must manually activate the partial high-beam distribution TL for about one second. In such conditions, the partial high-beam light distribution TL reflected from the dark slippery road FB of the highway is particularly good from the image capturing unit 2 detectable and from the image evaluation unit 3 evaluable, so that a particularly accurate position determination of the Teilfernlichtverteilung TL and thus a particularly accurate adjustment of the light units 4 of the vehicle 1 is possible. Furthermore, this involvement of the vehicle driver informs him in a particularly informative manner that an exact adjustment of the light units 4 his vehicle 1 is checked regularly and adjusted if necessary, so that the driver can be sure that an unnoticed malposition of the light units 4 and glare and thereby endanger other road users is excluded.

In addition to the already mentioned light distributions AL, LL, TL, ABL, ABL 'also enable further light distributions, which in particular by means of the light assist system of the vehicle 1 are used, such as the distribution of dangerous light, especially high-contrast bright-dark boundaries HDG and are therefore particularly simple in a comparable manner to the already described embodiments of the method of the image acquisition unit 2 detectable and from the image evaluation unit 3 evaluated with regard to the positioning of the light distributions, so that with these further light distributions, a corresponding adjustment of the light units 4 of the vehicle 1 is possible.

The image processing instructions [1], [2], [3] and [4] described in the various embodiments only provide examples of image processing in the image evaluation unit 3 to determine significant parts of the light distributions AL, LL, TL, ABL, ABL 'and to determine their position. In addition to these image processing rules, further classical image processing algorithms are applicable, by means of which specific image parameters are extracted and compared with a given light distribution AL, LL, TL, ABL, ABL 'or their given image parameters, in order to obtain an optionally present horizontal and / or vertical image Deviation of the current position of the light distribution AL, LL, TL, ABL, ABL 'from the predetermined target position of the light distribution AL, LL, TL, ABL, ABL' to determine and thus possibly the light units 4 by adjusting the corresponding means horizontally and / or vertically to adjust.

In the 16 and 17 a particularly advantageous embodiment of the method is shown. Here, first, as in 16 represented, an optical axis OA of the light units 4 pivoted from a starting position AS by a predetermined value VW in a lowered position ABS vertically. The optical axis OA is in 16 represented in the initial position AS as a dot-dash line and shown in the lowered position ABS as a solid line.

The starting position AS is a normal position of the light units in this example 4 in which the low-beam light distribution ABL or specific characteristics of the low-beam distribution ABL, for example the light-dark boundary HDG, at a relatively large distance of about 50 m to 60 m in front of the vehicle and with a relatively low contrast of a road surface of the road FR is reflected. The pivoting of the optical axis OA results in a shift of the low beam distribution ABL in the direction of the vehicle 1 , as in 17 shown. The low-beam light distribution ABL is shown here surrounded by a dot-dash line, a displacement resulting from the shift Abblendlichtverteilung ABL ', the bright-dark boundary HDG about five to ten meters in front of the vehicle 1 is shown surrounded by a solid line.

The optical axis OA is pivoted so far from the initial position AS as far down that the shifted low beam distribution ABL 'of the light units 4 as close as possible to the vehicle 1 , but still in a detection area EB of the image acquisition unit 2 reflected from the road surface. Limits of the detection range EB are in 16 represented by two dashed lines. In this way, the shifted low beam distribution ABL 'and in particular their specific features, in the example shown here in particular the cut-off line HDG, much more rich in contrast of the image capture unit 2 detectable and from the image evaluation unit 3 evaluable.

The vertical pivoting of the optical axis OA, for example, by means of the automatic headlamp leveling of the vehicle 1 performed on vehicles 1 with gas discharge lamps or LED bulbs anyway in the vehicle 1 is installed. This headlamp leveling is usually carried out by means of stepper motors, so that the beam range can be set very accurately and in this way also the optical axis OA of the light units 4 can be swiveled vertically downwards very precisely by the predetermined value VW. After carrying out the method, ie after checking the position of the shifted low beam distribution ABL 'and optionally after the automatic adjustment of the light units 4 , the optical axis OA is again swung back to the starting position AS, ie, vertically returned to the normal position, so that the light units 4 for a driving operation of the vehicle 1 are ready for use.

Therefore, the method is expediently carried out for safety reasons only if the vehicle 1 stands because, with shifted low-beam distribution ABL 'the environment of the vehicle 1 is lit only very limited. D. h., The vertical pivoting of the optical axis OA is only when the vehicle 1 carried out. When the vehicle is moving 1 This would endanger the vehicle occupants of the vehicle 1 and from other road users due to a lack of visibility of the driver.

The method preferably becomes immediately after an engine start and subsequent self-referencing of the light units 4 of the vehicle 1 carried out. Such self-referencing is used in vehicles 1 carried out, which have the automatic headlamp leveling to adjust a vehicle height and vehicle inclination corresponding headlamp range. Immediately thereafter, by means of the stepping motors of the headlamp leveling, the optical axis OA of the light units 4 pivoted vertically downwards by the predetermined value VW, the vertical and / or horizontal positions of the shifted low beam distribution ABL 'are extracted from that of the image acquisition unit 2 determined and from the image evaluation unit 3 evaluated image and compared with the predetermined target positions.

In this way the light units become 4 of the vehicle 1 checked at each ride and adjusted either automatically if necessary, or the driver is at least advised of a required adjustment. This will prolong driving with incorrectly adjusted light units 4 avoided, whereby a non-optimal brightness for the driver and dazzling other road users are avoided.

Appropriately, it is checked before determining the horizontal and / or vertical position of at least one predetermined part of the shifted low-beam light distribution ABL ', whether a relevant detection range of the image acquisition unit 2 is free of obstacles that would distort the determination of horizontal and / or vertical positions. The relevant detection area is in particular a part of the detection area EB of the image acquisition unit 2 in which the shifted low beam distribution ABL 'is detected.

If such obstacles are detected, no position determination of the shifted low beam distribution ABL 'is performed. If the shifted low-beam distribution ABL 'and, in particular, its specific features, ie the light-dark boundary HDG, are not reflected by the flat road surface, but by obstacles, then possibly incorrect positions are determined and thus a misadjustment of the light units 4 diagnosed, which is not real, however. This could lead to a misalignment of the light units 4 which is prevented by detecting the obstacles.

By the vertical lowering of the optical axis OA by the predetermined value VW, the shifted low-beam distribution ABL ', in particular its bright-dark boundary HDG, would have to be in a predetermined position in front of the vehicle and also at a predetermined position of the captured image. If the vertical and / or horizontal positions of the light-dark boundary HDG determined by the image evaluation of the acquired image in the image evaluation unit do not coincide with corresponding predetermined target positions, then either the information message is generated to the driver so that he can visit a workshop and the light units 4 can be adjusted, or it will be the Means for horizontal and / or vertical adjustment of the light units 4 controlled and the light units 4 are automatically adjusted in this way.

The in the 16 and 17 illustrated embodiment of the method can also be performed according to the other described light distributions AL, LL, TL, which have the characteristic features already described. In particular, the highway distribution LL is very well suited for this, since it has, as already described, the characteristic kink K of the light-dark boundary HDG at a predetermined position, so that with this highway distribution LL particularly well next to the vertical orientation, ie the beam range the light unit 4 also their horizontal orientation can be checked.

This is done by means of the light units 4 the highway distribution LL generates and by the vertical lowering of the optical axis OA in the direction of the vehicle 1 In this way, the light-dark boundary HDG, in particular its horizontal portion hA and its characteristic kink K can be particularly rich in contrast by means of the image acquisition unit 2 right in front of the vehicle 1 be recorded. If their vertical and / or horizontal positions determined in this way do not coincide with corresponding predefined setpoint positions, then here as well the information message is generated to the driver so that he can visit a workshop and the light units 4 or it can be the means for horizontal and / or vertical adjustment of the light units 4 controlled and the light units 4 are automatically adjusted in this way.

In 18 a flowchart of the inventive method for calibrating and / or adjusting not shown headlights of a vehicle is shown, wherein the headlights are provided for illuminating an in front of the vehicle vehicle environment. The headlamps include light bulbs, gas discharge lamps and / or light emitting diodes (= LED). In one embodiment of the headlights as LED headlights, the light-emitting diodes are preferably arranged in a so-called light-emitting diode array in a plurality of rows and columns.

The vehicle includes a detection unit (also not shown), by means of which images X of the vehicle surroundings in front of the vehicle and illuminated by the headlights are detected. The detection unit is a camera, in particular a camera of a night vision assistance system or a multifunction camera, for example for traffic sign detection or for the realization of a parking aid or an adaptive high-beam assistant.

If a current driving situation permits, the headlights are controlled in such a way that their pitch angle changes so that the scene is illuminated directly in front of the vehicle. This will be in the 22 and 23 shown light distributions LV of the headlights lowered so that in each case a minimum beam range LW is set, the headlight range LW in the 20 and 21 is shown in more detail.

Subsequently, by means of the detection unit, the images X of the vehicle surroundings situated in front of the vehicle and illuminated are detected and the recorded images are processed by means of an image processing method such that a light-dark boundary HDG of the light distributions LV of the headlights can be determined.

For this purpose, in a first method step S1, the captured images X are read out pixel by pixel and stored while maintaining a line information.

In a second method step S2, the stored images X are processed into overall result images Y in the image processing method, wherein light-dark transitions in the stored images X are amplified by means of at least one filter operator FO.

In a third method step S3, a position of a light-dark boundary HDG in a line structure of the overall result image Y is likewise determined in the image processing method, a distance being based on at least one determined image line in which the light-dark boundary HDG is located A of the light-dark boundary HDG is determined by the vehicle and a location L of the light-dark boundary HDG in the vehicle environment.

In a fourth method step S4, a deviation .DELTA.A, .DELTA.L of the distance A and the position L of the light-dark boundary HDG from a respective predetermined desired value A _Soll , L _{Soll is} determined in a calibration.

The method steps S1 to S4 are performed cyclically for different headlight settings.

By modeling a pitch angle drive, the actual headlamp angle is then determined and the headlamps are raised back to the prescribed level, i. H. the beam range LW is increased again to the prescribed size.

In a subsequent adjustment of the headlights in a fifth method step S5 parameters PAR of actuators of a headlight range control are set such that the deviation .DELTA.A, .DELTA.L of the distance A and the position L of the light-dark boundary HDG of the respective predetermined setpoint A _Soll , L _Should be eliminated automatically.

Since during the measuring process, ie during the determination of the deviation ΔA, ΔL of the distance A and the position L of the light-dark boundary HDG of the respective predetermined setpoint A _Soll , L _Soll an active headlamp setting is made certain security requirements must be met. Since a malposition of the headlights usually develops over a longer period, the determination is made at predetermined intervals. The determination is carried out, for example, analogous to service intervals of the vehicle after a minimum time interval, z. B. after about three months, or after a predetermined length of a covered distance, such. B. after 2500 km, wherein before the measurement is carried out automatically a measurement request is generated.

Alternatively or additionally, it is possible for sensors installed on the vehicle to be used in order to initiate a measurement request for an extraordinary measurement. This is the case, for example, when an acceleration sensor mounted on the vehicle detects a high transverse and / or longitudinal acceleration or when level sensors detect and signal rapid overrun of a bump, such as a brake hill. In order to enhance customer experience, it is also possible to manually start an extraordinary measurement by selecting a menu entry in a combination instrument or a so-called head unit of the vehicle.

The respective measurement request is stored until a situation occurs in which a stable measurement can be made. This situation includes in particular the absence of foreign traffic. The situation can be detected by means of a sensor of a cruise control or of an adaptive high-beam assistant. In addition, the adaptive high-beam assistant ensures that no external lighting, such as street lamps, is present.

Furthermore, by means of the night vision assistance system, a person and animal recognition is realized, so that dangers for the vehicle and objects located in the vicinity of the vehicle during the determination of the deviation ΔA, ΔL of the distance A and the position L of the light-dark boundary HDG of the respectively predetermined setpoint A _setpoint , L _setpoint should be avoided.

Since the lowering of the headlamps and the associated reduction of the headlamp range LW is accompanied by a reduction of a recognizable for the driver of the vehicle field of view, a maximum speed during the entire measurement is limited or the measurement is performed only at speeds of the vehicle below a predetermined maximum speed lie.

When legally prescribed headlamp setting the light-dark boundary HDG is located approximately at a distance A of 70 m to the vehicle. At a typical for a highway speed of 100 km / h, this distance in 2.5 s from the vehicle run over. This means that the vehicle reaches the position of the cut-off line HDG after 2.5 seconds. If the cut-off HDG is lowered to 30 m and the same duration of 2.5 s prescribed, the maximum speed is 43 km / h.

This maximum speed for the headlight measurement, d. H. the calibration of the headlights, is about a minimum speed for activating the adaptive high-beam assistant.

In addition, it will continue to ensure that the vehicle does not negotiate curves during calibration, as this will adversely affect safety during the measurement and its accuracy. This is realized with the aid of a navigation system of the vehicle, wherein the curves are determined from a digital map of the navigation system and a position of the vehicle is determined in particular by means of a GPS sensor.

If the described situation requirements are met, the recording of a measurement is started. For this purpose, the headlights are first individually or jointly controlled by driving the associated actuators of the headlamp level control via an on-board fieldbus such that the light distributions LV of the headlights are lowered and each a minimum beam range LW is set before performing the first process step S1. Thus, the light distributions LV in the captured image X are easily detectable. Any existing dynamic headlight range control is preferably deactivated during the measurement process and / or is put into a passive mode by a headlight control unit. This is done in particular when the vehicle is traveling at a virtually constant speed, since in this case the dynamic headlight range control is not required in the absence of road bumps.

To establish a lower limit for the headlamp lowering to minimize the beam range LW a prior knowledge of the accuracy of a previous headlamp setting is used. Alternatively or additionally, the lower limit is determined and determined in the image processing method.

Subsequently, images X of the vehicle surroundings in front of the vehicle are continuously taken by means of the detection unit installed on or in the vehicle. The recorded images X are stored with an associated current raw value of an adjustment of the headlight range control of the headlights and a raw value of the level sensors installed on the vehicle.

These stored images X, together with the raw data in the image processing method, are combined with an output of the image processing method so as to obtain a measurement or calibration of any misalignment of the headlights.

After taking pictures X with minimum headlight range LW, the headlight range LW of the headlights is restored to the previous level. In a particular application, the image evaluation takes place in such a way that the headlight range LW of the headlights is set to a corrected level immediately after the calibration. A period of such calibration and adjustment is, for example, 2.5 seconds when the headlamp headlamps LW of the headlamps are minimized. If the headlight range LW is reduced to a value above the minimum value, the time duration is reduced.

Furthermore, the period for the image acquisition is separated from the period for the calculation of the new calibration in order to minimize the time for the adjustment of the headlamps by means of the headlight range control.

By means of the image processing method, the position of the light-dark boundary HDG in the captured image X is determined for the headlights. Based on a back projection of pixel coordinates, the distance A of the light-dark boundary HDG to the vehicle or to the respective headlight is determined by means of a hole camera model and by means of calibration parameters of the detection unit.

19 shows a comparison between the thus determined distance A of the light-dark boundary HDG of the light distribution LV to a headlight of the vehicle and their image in the captured by the image capture unit image X. It is illustrated that a distance inaccuracies of 5 m in the range of distance A results from 30 m in a difference of 10 pixels px in the captured image X, and that a distance inaccuracy of 5 m in the range of 70 m only results in a difference of 1.1 pixels px in the captured image X. It follows that the possible measurement accuracy increases with decreasing distance A of the light-dark boundary HDG to the vehicle.

Prerequisite for the determination of the distance A and the position of the cut-off line HDG is that the vehicle and the light distributions generated by the headlights LV are on a flat surface.

From the comparison of a raw value control of actuators of the headlamp range control of the headlamps and measurements of the real headlamp range LW a linear relationship can be determined.

This linear relationship is in 20 shown, wherein the comparison between the raw value control of the actuators of the headlight range control and the measured light range LW for all four light modules of a headlight pair is shown. The headlights are designed as so-called PHB headlights, by means of which a partial high beam can be generated.

There is a linear relationship between the actuation of the servomotors, ie the parameter PAR of the servomotors and the luminous range LW for all four light modules. The four light modules are two pixel light modules PM1 and PM2, by means of which different light images can be generated, and two Basic light modules GM1 and GM2, which produce a basic light. Each headlamp includes one basic light and one pixel light module.

The illustrated straight line slopes are the same for both pixel light modules. The straight line slopes for both basic light modules are also the same. The straight lines GM1, GM2 represent the linear relationship between the control of the parameters PAR of the servomotors and the beam width LW for a left basic light module and a right basic light module and the straight lines PM1, PM2 for a left pixel light module and a right pixel light module of the headlights.

The line slope is assumed to be constant, the only relevant parameter for the headlight setting is the intercept. This is determined from the survey carried out from the recorded image X by solving a linear equation. Finally, in a headlight control unit, an accurate, in particular updated model for the relationship between the raw value control of the actuators of the headlight range control and the real headlight range LW deposited.

In addition, the measured values are particularly preferably processed by means of a temporal smoothing, for example by means of a low-pass filter, resulting in chronologically very robust measurement results.

21 shows a comparison between the measured headlight range LW and a headlamp leveling control of the right pixel light module. The crosses mark the measured values, the straight line PM2 is the regression line. An error between the regression line and the measured values is negligibly small, so that a linear model is particularly advantageously sufficient to describe this relationship.

In 22 is an image X recorded by means of the detection unit in the driving operation of the vehicle at an inclination angle of the headlamp of 4.5%, showing the light distribution LV in the vehicle environment in front of the vehicle.

23 shows an image X recorded by means of the detection unit while driving the vehicle at a tilt angle of the headlamp of X, which shows the light distribution LV in the vehicle environment in front of the vehicle. In this case, the light distribution LV is compared to 22 due to the increased inclination angle closer to the vehicle.

In 24 is shown in the second method step S2 in the image processing method from a stored image X further processed overall result image Y. The overall result image Y was determined by means of an artificial neural network and applied to the recorded image X by means of a convolution. In the range of a target class of the light-dark border HDG of the headlights the answer is strongly pronounced. False alarms can still be recognized on the right-hand side of the road, which are suppressed with the help of a larger learning sample.

By means of the image processing method, the light-dark boundary HDG of both headlights is localized in the overall result image Y. The light distribution LV exhibits characteristic properties in the recorded image X, which are determined on the basis of image processing operators adapted thereto, i. H. the filter operators FO are recognized. In this case, light-dark transitions in the stored images X are amplified by means of the filter operators FO and focused on a maximum of two image lines.

Here, a distinction is made between filter operators FO optimized on the image X and filter parameters FO produced by hand and parameterized.

The filter operators FO are determined prior to implementation in the image processing method in an optimization method or in a test method, wherein the respectively associated filter operator FO is determined in the optimization method and the test method for different real illumination scenarios.

In a first exemplary embodiment of the image processing method, a bottommost pixel line of a freely programmable LED headlight is detected.

For this purpose, a simple, on image sections of the size 15 × 15 pixels px working model is calculated, which is able to distinguish the image sections of the associated cut-off line HDG from all other image sections. The model is the result of an optimization problem, which is solved with the help of an artificial neural network.

As image preprocessing, a mean-variance normalization of the image sections of the captured image X is used. The probability image calculated according to this model then results as follows:
On the basis of a mean-variance-normalized convolution of the recorded image X with some prototypical image sections, several convolution images are determined. These convolution images are further processed by a nonlinear transfer function, and then an entry-wise linear combination is formed. Further evaluation of a nonlinear transfer function yields the final and represented overall result image Y.

An additional temporal filtering was not applied in the illustrated embodiment.

In the overall result image Y, there are regions with high entries if and only if this region is highly likely to belong to the target class, ie. H. if the respective 15 × 15 pixel image sections are likely to belong to the light / dark boundary HDG of the lowest pixel line of the headlight.

With simple subsequent image processing operators, such as forming a row sum vector of a probability image and finding a cluster point in that vector, the light-dark boundary HDG is accurately located. In this case, the image line in the overall result image is determined in which the light-dark boundary HDG is located and the distance A of the light-dark boundary HDG from the vehicle and the location L of the light-dark boundary HDG in the vehicle environment are off Derived image line derived.

The advantage of this method is its high quality ruggedness and independence from interference, such. For example, heavily textured or contaminated pavement, in the captured images X. By extending the learning sample for the model, the robustness can be further increased.

In an alternative embodiment, a manually created and parameterized filter operator FO is used. The filter operator FO is for example a barycentric edge operator, by means of which a gradient in a small image area is approximated. Alternatively or additionally, other filter operators FO can also be used for approximation of the gradient. This includes, for example, the so-called Sobel operator.

The lower limit of the light distribution LV is characterized in the recorded image X by a vertical brightness increase from bottom to top. In order to find the lower limit of the spotlight beam, image areas in which the gradient is as accurate as possible in that direction are marked in a total result image Y with a value close to one, and other areas are given values close to zero.

By means of this feature alone, however, no robust marking of the sought area is feasible because z. B. in relatively homogeneous areas in the image, the gradient often shows vertically upwards, although hardly a difference in brightness. In these areas, however, it is noticeable that the gradients of adjacent environments often point in completely different directions.

Therefore, the local variance is calculated for all image sections of a certain size. In doing so, another overall result image Y is created in which areas of low variance are marked with a value close to one and areas of high variance receive a value close to zero.

This further overall result image Y is multiplied pixel by pixel with the previously determined overall result image Y, as a result of which the lower edge of the light distribution of the headlight is marked very clearly. Any occurring disturbances are filtered out by a temporal averaging of several total result images Y. In turn, the image line in the overall result image Y is determined, in which the light-dark boundary HDG is located, and the distance A of the light-dark boundary HDG from the vehicle and a location L of the light-dark boundary HDG in the vehicle environment determined from the position of the image line in the overall result image Y.

An overall result image Y of the barycentric edge operator averaged over five images is in 25 shown. A large part of a positive filter response is in the range of the light-dark limit HDG of the light distributions LV of the headlights. By applying a region-of-interest for a left and a right headlight and simple evaluation methods, such as line sums or a Hough transform for straight lines in the regions-of-interest, the exact headlight inclination angles from a geometry of the detection unit can be calculated in a particularly robust manner ,

This approach is used in a further embodiment by further operations, the z. B. consider a length of the gradient, extended. The individual overall result images Y are preferably weighted differently depending on the informative value of the examined feature. In a particularly advantageous manner, model knowledge about the headlight type used is used to specify the measurement.

With an inserted, freely programmable LED headlight, an opening angle of a pixel is 1 °. Finding the bottom and top edges of the bottom pixel row increases the likelihood that the found areas are really the searched bottom pixel row.

in das von der Erfassungseinheit aufgenommene Bild umwandeln lässt. Das Ziel ist es nun, eine Abbildungsvorschrift zu finden, die ein normiertes Bild, unabhängig von den konkreten Werten a und b, berechnet. Im Idealfall ergibt sich damit das latente oder ein dazu äquivalentes Bild.With the help of an image model, an invariance to disturbances, such. B. a camera control, different optics and sensor sensitivities achieved. The image model includes pixel-based affine-linear transformations with positive scaling. That is to say, it is assumed that a latent image is obtained with pixel-by-pixel application of a mapping rule x ↦ i ax + b with arbitrary a> 0 and

can be converted into the captured by the capture unit image. The aim now is to find a mapping rule that calculates a normalized image, regardless of the specific values a and b. Ideally, this results in the latent or an equivalent image.

diese Anforderungen in statistisch robuster Weise, denn für einen beliebigen Bildausschnitts

ergibt sich f(aD + b) zu f(D) für alle a > 0 und

In diesem Fall ist f(D) der normierte Bildausschnitt, der unabhängig von der Belegung von a und b stets die gleichen Eigenschaften aufweist.Obviously fulfilled

these requirements in a statistically robust manner, because for any image section

f (aD + b) results in f (D) for all a> 0 and

In this case, f (D) is the normalized image detail, which always has the same properties regardless of the assignment of a and b.

übertragen. Hierfür sei

und

Let x ₁ , ..., x _{n be} an arbitrary finite sequence of numbers. Their mean value μ results in (x ₁ + ... + x _n ) / n. Its variance σ ² results in [(x ₁ - μ) ² + ... + (x _n - μ) ² ] / n. With the help of the shift theorem, the variance becomes [(x ₁ ) ² + ... + (x) ² ] / n - 1 ² . In the following, these considerations are applied to two-dimensional image sections of the size u × v pixels px of an image

transfer. For this purpose be

and

ein beliebiger Bildausschnitt, ergibt sich der Mittelwert der Pixel in D durch Multiplikation der Einträge von D mit den zugehörigen Einträgen in J und der anschließenden Aufsummierung aller Produkte. Mit Hilfe des Faltungsoperators „*” lässt sich dieser Vorgang für alle Bildausschnitte D eines Bildes X ausdrücken.Is now

an arbitrary image section results in the average of the pixels in D by multiplying the entries of D by the corresponding entries in J and the subsequent summation of all products. With the help of the convolution operator "*", this process can be expressed for all image sections D of an image X.

und gibt für jedes Pixel px im Bild X einen Mittelwert in einer u × v-Nachbarschaftsumgebung an.A matrix μ results to

and for each pixel px in the image X indicates an average in a uxv neighborhood environment.

die Varianz in einer u × v-Nachbarschaftsumgebung an. Hierbei bezeichnet „∘” das so genannte Hadamard-Produkt, d. h. des eintragsweise Produkt von Matrizen.Because of the shift rate gives

the variance in a uxv neighborhood environment. Here, "∘" denotes the so-called Hadamard product, ie the eintragsweise product of matrices.

This results in the following relationships: (X ∘ X) · J ≙ E (X ² ) [10] and (μ ∘ μ) ≙ E (X) ² . [11]

ist σ die Standardabweichung für jedes Pixel px in einer u × v-Nachbarschaftsumgebung, die Wurzel wird eintragsweise genommen.According to

For example, if σ is the standard deviation for each pixel px in a uxv neighborhood environment, the root is taken in by entry.

angewendet wird. Dieses Verfahren lässt sich für das gesamte Bild X mit Hilfe der Faltungsalgebra wie folgt ausdrücken:

In order to realize a reduction in the computational effort for normalization and convolution, each possible image detail D is normalized from the recorded image X of size u × v average and standard deviation, before the filter operator FO with the filter kernel

is applied. This method can be expressed for the entire image X using the convolution algebra as follows:

The filter kernel W is here the result of an optimization problem. The filter kernel W is adapted in such a way that there is always a high entry in the overall result image Y, if it is an image section from the cut-off line HDG. A low entry is at the positions in the overall result image Y, which show no overlap with the light-dark boundary HDG.

This is performed for a plurality of different convolution kernels W ₁ ... W _n , so that a plurality of result images Y ₁ ... Y _n results. Applying a nonlinear transfer function f yields result-based images Z _i calculated _in accordance with the invention Z _i : = f (Y _i ) [14] for all i, where i is a count variable. The result images Z _i are then combined linearly and combined to form an overall result image Y.

In other words: To reinforce the light-dark transitions, a plurality of filter operators FO is used in the stored images X in the second method step S2, whereby a first result image is determined by means of the filter operators FO and then by means of a non-linear transfer function first result image is converted into a second result image. Within the respective first result image, pixel contents below a predetermined brightness threshold value are suppressed disproportionately and pixel contents above the predefined brightness threshold value are increased disproportionately, with the second result images being combined to form the overall result image Y. In particular by means of a set of up to 16 filter operators FO, the bright-dark boundaries HDG are particularly reliable and precisely determinable.

LIST OF REFERENCE NUMBERS

AV

low beam

BA1

image section

BA2

image section

B1

first reference picture

B2

second reference picture

e

distance

f

accordance

FV

Beam light distribution

GB

filtered image

HDG

Light-off

KB

correlation image

POS _B

position

POS _g

global position

POS _l

local position

α

tilt angle

1

vehicle

2

Image capture unit

3

image evaluation

4

light unit

AL

Motorway light distribution

ABL

low beam

ABL

shifted dipped beam distribution

SECTION

lowered position

AS

starting position

BPS

column

BPZ

image line

bU

illuminated environment area

D

difference image

EB

detection range

F

bright spot

FB

roadway

Ha

horizontal part of the cut-off line

hB

bright area

hL

horizontal line

hSK

horizontal shadow edge

K

Kink the light-dark border

LL

Country light distribution

nB

not illuminated area

O

object

OA

optical axis

S

main emphasis

TL

Part beam distribution

vL

vertical line

vSKl

left vertical shadow edge

vSKr

right vertical shadow edge

VW

predetermined value

W

wall

A

distance

A _target

setpoint

FO

filter operator

GM1, GM2

Just

L

location

L _Soll

setpoint

LV

light distribution

LW

Lighting range

PAR

parameter

PM1, PM2

Just

px

pixel

S1 to S5

step

X

image

Y

Overall result image

.DELTA.A

deviation

.DELTA.L

deviation

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 19602005 A1 [0003]
• DE 4341409 A1 [0004]
• DE 102007049619 A1 [0005]
• DE 19902015 A1 [0006]
• DE 4122531 A1 [0007]
• US 5633710 A1 [0008]
• DE 102010048689 A1 [0009]
• DE 102009047932 A1 [0010]
• DE 102005040980 A1 [0011]

Claims (10)

Method for adjusting at least one headlight and / or a light unit ( 4 ) of a vehicle ( 1 ), wherein by means of a detection unit, in particular an image acquisition unit ( 2 ) at least one image (X) of one in front of the vehicle ( 1 ) and by means of the at least one headlight or the light unit ( 4 ) illuminated vehicle environment is detected, and that from the at least one image (X) while maintaining a row and / or column information, the position of a cut-off (HDG) determined and compared with a desired position (A _Soll , L _Soll ) in which, based on the determined deviation (ΔA, ΔL) between the position and the desired position, the adjustment of the headlights or the light unit (FIG. 4 ) is changed, characterized in that the contrast ratio of the bright-dark boundary (HDG) is changed. A method according to claim 1, characterized in that the contrast ratio is changed by superposition of several images, by switching between different light distributions or by image processing by means of filter operators (FO). Method according to claim 1 or 2, characterized in that the captured images (X) are further processed into overall result images (Y), wherein light-dark transitions in the stored images (X) are amplified by means of at least one filter operator (FO), a position (L) of a light-dark boundary (HDG) by means of the headlight and / or the light unit ( 4 ) is determined in a line structure of the overall result image (Y), wherein on the basis of at least one determined image line in which the light-dark boundary (HDG) is located, a distance (A) of the light-dark Frontier (HDG) of the vehicle (HDG) 1 ) and / or a location (L) of the light-dark boundary (HDG) in the vehicle environment is determined A method according to claim 2 or 3, characterized in that the cut-off line is focused by means of the filter operator (FO) to a maximum of two picture lines. A method according to claim 2, characterized in that by means of the filter operators (FO) in each case a first result image is determined and then by means of a non-linear transfer function, the respective first result image is converted into a second result image, wherein within the respective first result image pixel contents below a predetermined brightness threshold is disproportionately suppressed and pixel contents are increased disproportionately above the predetermined brightness threshold, the second result images are combined to the overall result image (Y). A method according to claim 1 or 2, characterized in that a first image of the highway light distribution (AL) is detected, in the highway light distribution (LL) switched and a second image is detected and switched back to the motorway light distribution (AL) and detects a third image is, whereby by means of the image evaluation unit ( 3 ) an average image of the first and third image is formed and a difference image (D) of the mean value and the second image is determined. Method according to one of claims 1 or 2, characterized in that by means of the image capture unit ( 2 ) a first reference image (B1) of a low beam distribution (AV) and a second reference image (B2) of a high beam distribution (FV) are detected, wherein the first reference image (B1) immediately before switching and the second reference image ( 82 ) are detected immediately after the switching between the low beam distribution (AV) and the high beam distribution (FV) and from the reference images (B1, B2) a common correlation image (KB) is generated, and that the contrast ratios in an image section (BA1) of the correlation image (BA1) KB), wherein first a local position (POS ₁ ) of the light-dark boundary (HDG) in the image section (BA1) is determined and in dependence of a position (POS _B ) of the image section (BA1) in the correlation image (KB ) the global position (POS _g ) of the light-dark boundary (HDG) in the correlation image (KB) is determined. Method according to one of the preceding claims, characterized in that prior to adjustment of the headlight and / or the light unit ( 4 ) the light distribution (LV) is lowered so that a minimum light range (LW) is set. Method according to one of the preceding claims, characterized in that it is checked whether a relevant detection area (EB) of the image acquisition unit ( 2 ) is free of obstacles. Method according to one of the preceding claims, characterized in that prior to capturing the at least one image, an optical axis (OA) of the at least one light unit ( 4 ) and / or the headlight from a starting position (AS) by a predetermined value (VW) is pivoted vertically.

Images