Method and device for calibrating extrinsic parameters of an environmental sensing camera

Abstract

The invention relates to a method for calibrating extrinsic parameters of an environment sensing camera (1) of a vehicle (2), wherein an original camera image (OB) of the vehicle's (2) environment is captured by means of the environment sensing camera (1). In this process, a deviation of the original camera image (OB) from the synthetic camera image (Bs) generated from a satellite image (SB) is determined by comparing the original camera image (OB) with the synthetic camera image (Bs), whereby an existing calibration of the extrinsic parameters is checked for correctness depending on the magnitude of the deviation. The invention further relates to a device for calibrating extrinsic parameters of an environment sensing camera (1) of a vehicle (2) and a method for operating a vehicle (2).

Description

The invention relates to a method for calibrating extrinsic parameters of an environmental sensing camera of a vehicle according to the preamble of claim 1. The invention further relates to a device for calibrating extrinsic parameters of an environmental sensing camera of a vehicle according to the preamble of claim 10. Furthermore, the invention relates to a method for operating a vehicle. The operation of vehicle driver assistance systems generally relies on data acquired by a variety of different sensors. For driver assistance while driving, including during parking maneuvers, image data captured by surround-view cameras is particularly essential. These cameras can perform two-dimensional object detection and classification, as well as resolve the three-dimensional positions of objects such as pedestrians, other vehicles, etc. Such functionality requires precise intrinsic and extrinsic calibration of the camera relative to a vehicle coordinate system. Intrinsic calibration involves determining parameters such as an internal calibration matrix K, focal length, and optical axis.Extrinsic calibration describes the external position of the environmental detection camera in a defined coordinate system. The coordinate system of a vehicle's surround-view camera and the vehicle's own coordinate system have a defined relationship, allowing the coordinates of an object detected by the surround-view camera to be transformed into coordinates within the vehicle's coordinate system. If the camera's positioning is inaccurate or misaligned—that is, if the calibration of extrinsic parameters, particularly translation and orientation, is incorrect—then the position of the object detected by the surround-view camera will be displayed incorrectly in the vehicle's coordinate system, or alternatively, in a shared coordinate system of the surround-view camera and the vehicle. This significantly impairs the functionality of driver assistance systems, such as emergency braking assist, automated parking, etc.There is a risk that a driver assistance system operating based on image data of the detected object captured by the surround-view camera will not function correctly, and thus the vehicle will not avoid the object or warn of an impending collision, potentially resulting in collisions between the vehicle and the object. Calibration ensures the proper functioning of the systems processing the camera data. It is generally known from the state of the art that when calibrating environmental detection cameras of vehicles in the context of driver assistance, a distinction is made between calibration in the production of the vehicle and calibration during a driving operation of the vehicle at a customer's premises as part of an online calibration. During vehicle production, a static calibration is generally performed in a production area. This involves calibrating the vehicle's sensors over the course of the production process. For example, a Zhang or Tsai calibration is used, in which target objects, also known as targets, are positioned in different locations with known dimensions within a controlled environment. These targets exhibit defined patterns, such as a checkerboard pattern. A vehicle's environmental sensing camera, which is being calibrated, detects these known patterns and generates a two-dimensional image of the pattern. This 3D-to-2D mapping is also known as homography.An optimization problem can now determine the position of the environmental sensing camera relative to the corresponding target and to a vehicle origin, for example, the center point of a rear axle. This extrinsically calibrates the environmental sensing camera. Intrinsic calibration is usually performed by the manufacturer of the environmental sensing camera. During vehicle operation, calibration information determined in the aforementioned static calibration can repeatedly change. This can result, for example, from a displacement of a surround-view camera mounted in a folding and / or extendable exterior mirror when the mirrors are folded or extended, from strong vibrations, etc. This necessitates recalibrating extrinsic calibration parameters during vehicle operation, either manually or online. Such online calibration can be performed, for example, using camera-based segmentation. Objects of known size are segmented. Once the distance between these segments is known, extrinsic calibration can be performed.It is also known to perform tracking of characteristic features over time in a moving vehicle, a process also known as bundle adjustment. However, this does not allow for precise calibration when the vehicle is stationary or traveling at low speeds. From EP 1 550 982 B1, a device and a method for calibrating an image sensor system arranged in a motor vehicle are known, wherein the image sensor system is calibrated as a function of image signals from the image sensor system of a calibration object located in the road plane. The image sensor system is calibrated as a function of a first image signal and a second image signal. The first image signal is generated at a first position of the image sensor system relative to the calibration object, and the second image signal is generated at a second position, the second position being different from the first position. By moving the motor vehicle forward or backward from the first position of the image sensor system, the second position of the image sensor system is reached.The image signals and vehicle movement data between the first and second positions are used to calculate the installation parameters of the image sensor system and for calibration. The vehicle movement data includes signals from a steering angle sensor. Furthermore, Zuoyue Li, et al.: “Sat2Vid: Street-view Panoramic Video Synthesis from a Single Satellite Image”; arXiv:2012.06628v3 [cs.CV], May 6, 2021, discloses a method for synthesizing temporally and geometrically consistent street-view panoramic videos from a single satellite image and camera trajectory. For geometric and temporal consistency, a 3D point cloud representation of a scene is created, and dense 3D-2D correspondences, reflecting the geometric scene configuration derived from the satellite view, are maintained across frames. For synthesis in 3D space, a cascaded network architecture with two hourglass modules is implemented to generate pointwise coarse and fine features from semantics and latent vectors per class, followed by projection onto frames and an upsampling module to obtain a final realistic video. From the article titled "Image stitching" in Wikipedia, The Free Encyclopedia, November 5, 2024. URL: https: / / en.wikipedia.org / w / index.php?title=lmage.stitching=1255591740 [accessed January 28, 2026], a general method for stitching images is known in which several photographic images with overlapping fields of view are combined to form a panorama or a high-resolution image. The process typically involves image registration, calibration, and blending of the images to achieve seamless results. German patent DE 10 2022 001 120 A1 discloses a method for calibrating a vehicle camera. In this method, a digital surface model is provided in three-dimensional space for a determined vehicle position. Features are extracted from an image of the vehicle's surroundings captured by the camera and projected into the three-dimensional space. Subsequently, intersection points between the projected features and corresponding features from the digital surface model are determined. These intersection points are transferred into a two-dimensional space of the camera, and the similarity between these intersection points and the original camera image is used as a measure for calibration. The invention is based on the objective of providing a novel method and a novel device for calibrating extrinsic parameters of an environment detection camera of a vehicle, as well as a novel method for operating a vehicle. The problem is solved according to the invention by: - ​​a method for calibrating extrinsic parameters of an environment detection camera of a vehicle, which has the features specified in claim 1, - a device for calibrating extrinsic parameters of an environment detection camera of a vehicle, which has the features specified in claim 10, and - a method for operating a vehicle, which has the features specified in claim 11. Advantageous embodiments of the invention are the subject of the dependent claims. In a procedure for calibrating extrinsic parameters of a vehicle's environment detection camera, an original camera image of the vehicle's environment is captured using the environment detection camera. According to the procedure, it is provided that: - a synthetic camera image generated from a virtual camera perspective of the environment detection camera is requested and received from a central computing unit based on a satellite image of the vehicle's surroundings; - the synthetic camera image is compared with the original camera image, and in the comparison, a deviation of the original camera image from the synthetic camera image is determined; and - a measure describing the deviation is transformed from the satellite coordinate system into the vehicle coordinate system, and an existing calibration of the extrinsic parameters is checked for correctness using the transformed measure. Known methods for calibrating cameras during vehicle operation, such as vanishing point and vanishing line methods, Manhattan geometry, camera overlap, and secondary-based calibration methods using radar, lidar, etc., exhibit limitations in defined functional areas. For example, Manhattan geometry methods are impaired when driving in rural areas without the necessary infrastructure. The central processing unit is designed to receive satellite images of the vehicle's surroundings from a satellite and generate synthetic camera images of the vehicle's environment using one of the aforementioned methods. The processing unit can also be integrated into the vehicle, or at least partially integrated into it, i.e., implemented as a vehicle component. The present method advantageously compensates for certain weaknesses of the aforementioned methods. Specifically, it enables the calibration of extrinsic parameters of a vehicle's environmental sensing camera during vehicle operation—that is, online calibration—with reference to 3D vehicle sensors. For example, this method can be used redundantly with existing calibration methods. In contrast to prior art methods, this method does not require long distances traveled by the vehicle to precisely determine the extrinsic parameters. The present method can be applied regardless of the type of environmental sensing camera. The environmental sensing camera can be a telephoto camera, a fisheye camera, a surround-view system, a wide-angle camera for capturing the area in front of the vehicle (also known as a front wide camera, or FWC), or another type of camera. The method advantageously allows for the recurring evaluation of current calibration information. The method is particularly advantageous because it functions globally and is independent of prevailing weather conditions and the specific scenario in the vehicle's environment. In particular, the method does not require any specific image features, such as orthogonal edges in the captured original images. Furthermore, no input from the vehicle's driver is required to execute the method. According to the invention, it is provided that: - a geographical actual vehicle position and an actual vehicle orientation are determined; - from the actual vehicle position, a virtual camera position and a virtual camera orientation are generated in the vehicle coordinate system by means of translation to a camera position of the environment detection camera, and from the actual vehicle orientation, a virtual camera position and a virtual camera orientation are generated in the vehicle coordinate system by means of rotation to a camera orientation of the environment detection camera, for example by means of so-called vector repositioning; - the virtual camera position and the virtual camera orientation are transformed into the satellite coordinate system; and - the received synthetic camera image, which is generated in the satellite coordinate system from the at least one satellite image of the vehicle's environment (2) according to the virtual camera perspective formed by the virtual camera position and the virtual camera orientation,is requested and received from the central computing unit. This design enables particularly simple and precise calibration of the environmental sensing camera. In addition to the environmental sensing camera and a position sensing unit for determining the vehicle's current position and orientation (e.g., GPS sensors), no further vehicle-integrated sensors are required to carry out the procedure. Position sensing units, especially as components of navigation systems, are already installed in the majority of modern vehicles. This means the procedure can be implemented with minimal hardware and using established and reliable technology. In another possible embodiment of the method, the comparison is carried out by means of pixel registration, wherein - at least one pixel is determined in the synthetic camera image in which the original camera image has the highest agreement with the synthetic camera image, and - as a measure of the deviation between the pixel and a corresponding pixel of the original camera image, an image vector is determined in the satellite coordinate system. Using such pixel registration, a deviation of the original camera image from the synthetic camera image can be determined particularly accurately and reliably. In another possible embodiment of the procedure, it is provided that several synthetic single-camera images from different virtual single-image camera perspectives are requested and received from the central computing unit in the satellite coordinate system from at least one satellite image, and that the synthetic camera image is generated from the synthetic single-camera images using a stitching algorithm. This allows for a comprehensive representation of the vehicle's surroundings in the synthetic camera image. In another possible embodiment of the method, the virtual single-image camera perspectives are selected based on the expected maximum calibration error of the environmental sensing camera. This allows the display area of ​​the synthetic camera image to be adapted to the expected maximum calibration error. This ensures that the deviation between the original camera image and the synthetic camera image can be reliably determined even with large calibration errors. In another possible embodiment of the method, the virtual single-image camera perspectives are chosen in such a way that the synthetic single-image camera pictures overlap. This makes it possible to create the synthetic camera picture from the synthetic single-image camera pictures without any gaps. In another possible embodiment of the method, the calibration of the extrinsic parameters determines at least one deviation of the translation of the environmental perception camera from a target translation and one deviation of the orientation of the environmental perception camera from a target orientation. Using these parameters, the camera position and camera orientation in space can be precisely described. In another possible embodiment of the procedure, if an incorrect existing calibration is detected, the existing calibration is replaced by a calibration derived from the transformed measurement. This enables a correction and update of the calibration, so that image data captured by the environmental perception camera can be reliably adjusted with the correct calibration information, and consequently, driver assistance functions based on the captured image data can be executed with high functionality. In another possible embodiment of the procedure, it is provided that if the transformed measure falls below a predetermined threshold, the existing calibration is determined to be correct, and if the transformed measure exceeds a predetermined threshold, the existing calibration is determined to be incorrect. By setting at least one such predetermined threshold, it is possible to control in a simple and reliable way what level a calibration error of the environmental sensing unit must reach before it is corrected. In another possible embodiment of the procedure, calibration is performed cyclically and / or at intervals and / or based on events while the vehicle is in operation. This allows for regular and needs-based verification and, if necessary, correction of the calibration of the environmental perception camera. In a device for calibrating extrinsic parameters of a vehicle's environmental sensing camera, the environmental sensing camera is configured to capture original camera images of the vehicle's environment. The device comprises: - a data processing unit, which is configured to request and receive a synthetic camera image generated from a satellite image of a vehicle's environment by a central computing unit, from a virtual camera perspective of the environment detection camera; - compare the synthetic camera image with the original camera image and determine any deviation of the original camera image from the synthetic camera image in the comparison; and - transform a measure describing the deviation from the satellite coordinate system into the vehicle coordinate system and use the transformed measure to check the correctness of an existing calibration of the extrinsic parameters. The device offers the same advantages as the previously described method. In one embodiment of the device according to the invention, it comprises: - a position detection unit configured to determine a geographic actual vehicle position and an actual vehicle orientation, - wherein the data processing unit is configured to generate a virtual camera position and a virtual camera orientation in the vehicle coordinate system from the actual vehicle position by means of displacement to a camera position of the environment detection camera and from the actual vehicle orientation by means of rotation to a camera orientation of the environment detection camera, - transforming the virtual camera position and the virtual camera orientation into the satellite coordinate system and In the satellite coordinate system, a synthetic camera image generated from at least one satellite image of the vehicle's surroundings, which is created according to the virtual camera perspective formed from the virtual camera position and the virtual camera orientation, is to be requested and received from a central computing unit. In the inventive method for operating a vehicle, the existing calibration of extrinsic parameters of the vehicle's environment detection camera is checked for correctness by means of a aforementioned method, and if an incorrect existing calibration is detected, at least one function of at least one driver assistance system of the vehicle, which is executed depending on image data captured by means of the environment detection camera, is deactivated. This allows malfunctions or incorrect decisions of the driver assistance system to be effectively avoided, and enables a driving task performed by the driver assistance system to be handed over to the vehicle's driver. This increases the road safety of the vehicle and other road users in its vicinity. Exemplary embodiments of the invention are explained in more detail below with reference to drawings. Figure 1 schematically shows a process for calibrating extrinsic parameters of a vehicle's environmental sensing camera. Figure 2 schematically visualizes a first step of the process according to Figure 1. Figure 3 schematically visualizes a second step of the process according to Figure 1. Figure 4 schematically visualizes a third step of the process according to Figure 1. Figure 5 schematically visualizes a fourth step of the process according to Figure 1. Figure 6 schematically visualizes a fifth step of the process according to Figure 1. Figure 7 schematically visualizes a sixth step of the process according to Figure 1. Figure 8 schematically visualizes a seventh step of the process according to Figure 1. Figure 9 schematically visualizes an eighth step of the process according to Figure 1.10 schematically a visualization of a ninth process step of the process according to Fig. 1 . Corresponding parts are marked with the same reference symbols in all figures. Figure 1 shows a possible embodiment of a method for calibrating extrinsic parameters of an environmental sensing camera 1 of a vehicle 2. Figures 2, 3, 4, 5, 6, 7, 8, 9 to 10 show visualizations of process steps VS1 to VS9 of the method according to Figure 1. The method is used to calibrate the extrinsic parameters of the environmental sensing camera 1 during a color drive of the vehicle 2 in order to verify or, if necessary, correct existing calibration information. The method is independent of the type of environmental sensing camera 1 to be calibrated. In the illustrated embodiment, the method is described using an environmental sensing camera 1 designed as a wide-angle camera for capturing the area in front of the vehicle 2. Alternatively, the environmental sensing camera 1 can also be any other type of camera, for example, a telephoto camera or a so-called fisheye camera of a surround-view system of the vehicle 2. In a first process step VS1, visualized in Fig. 2, the environmental sensing camera 1 is initially calibrated, for example, during a static calibration performed during the production of the vehicle 2 or during an online calibration while the vehicle 2 is in operation. During this calibration, intrinsic and extrinsic calibration information is determined. This involves referencing a camera coordinate system KK of the environmental sensing camera 1 to a vehicle coordinate system FK. The camera coordinate system KK has a longitudinal axis xK, a transverse axis yK, and a vertical axis zKauf. The vehicle coordinate system FK originates, for example, at the center point of a front axle 2.1 of the vehicle 1 and also has a longitudinal axis xF, a transverse axis yF, and a vertical axis zFauf.After this initial calibration, the extrinsic parameters of the environment detection camera 1 are a camera position POSK and a camera orientation ARK in the form of a rotation and translation in relation to the vehicle coordinate system FK, which is shown in more detail in Fig. 5. The subsequent procedure steps VS2 to VS10 described below are performed to determine, validate, and / or adjust the extrinsic parameters of the environmental sensing camera 1, as initially determined or as determined during subsequent operation of the vehicle 2. However, performing procedure step VS1 before procedures VS2 to VS10 is not mandatory. Alternatively, initial calibration can also be performed using the procedure steps VS2 to VS10 described below. In a second process step VS2, visualized in Fig. 3, an original camera image OB of the vehicle's surroundings is captured using the environmental sensing camera 1. The original camera image OB shown in the upper part of Fig. 3 is, for example, a camera image captured by the environmental sensing camera 1, which is configured as a wide-angle camera to capture an area in front of the vehicle 2. The original camera image OB shown in the lower part of Fig. 3 is, for example, a camera image of an area in front of the vehicle 2, captured by the fisheye camera of the vehicle 2's surround view system. Subsequently, in a third process step VS3, visualized in Fig. 4, a geographic actual vehicle position POSist and an actual vehicle orientation ARister are determined using a position acquisition unit. In particular, this determination is carried out using a satellite-based positioning system of the vehicle. Subsequently, in a fourth process step VS4, visualized in Fig. 5, the actual geographic vehicle position POSist and the actual vehicle orientation ARist are modified, for example by means of so-called vector repositioning, so that they correspond to the calibrated extrinsic parameters, i.e., the camera position POSK and camera orientation ARK, of the environmental sensing camera 1. Here, a virtual camera position POSV and a virtual camera orientation ARvi, shown in more detail in Fig. 6, are generated in the vehicle coordinate system FK from the actual vehicle position POSist by translation to the calibrated camera position POSK of the environmental sensing camera 1, and from the actual vehicle orientation ARist by rotation to the calibrated camera orientation ARK of the environmental sensing camera 1. Subsequently, in a fifth process step VS5, visualized in Fig. 6, the virtual camera position POSV and the virtual camera orientation AR are transformed into a satellite coordinate system, for example, the so-called World Geodetic System 1984 (WGS 84), using the data processing unit. Furthermore, a satellite image SB of the vehicle 2's surroundings at the virtual camera position POSV is requested from a central processing unit (not shown) using at least one image acquisition unit of a satellite. The central processing unit is, for example, a server of a service that retrieves satellite images SB at the current vehicle position from satellites. Alternatively, the central processing unit can also be implemented as a component of the vehicle. Subsequently, using the virtual camera position POSV and the virtual camera orientation ARv, a two-dimensional synthetic camera image Bs of the vehicle's surroundings is generated according to a so-called Sat2Cam imaging or a so-called Synthetic Street View algorithm, as described, for example, in Zuoyue Li, et al.: "Sat2Vid: Street-view Panoramic Video Synthesis from a Single Satellite Image"; arXiv:2012.06628v3 [cs.CV], May 6, 2021. This image is shown in more detail in Fig. 8. These camera images Bs are determined by the central processing unit (not shown) from the satellite image(s) from the perspective of the virtual camera position POSV and in the direction of the virtual camera orientation ARv, and are requested and received by the data processing unit. For this purpose, in a sixth process step VS6, visualized in Fig. 7, several synthetic single-camera images EBs1 to EBsn, in particular an array of synthetic single-camera images EBs1 to EBsn, from different virtual single-image camera perspectives in the satellite coordinate system SB are requested from the central processing unit. An offset mapped by the different virtual single-image camera perspectives is selected depending on an expected maximum calibration error of the environmental sensing camera 1, wherein the expected maximum calibration error is known from applications or a design phase of the environmental sensing camera 1. Subsequently, in a seventh process step VS7, visualized in Fig. 8, the synthetic camera image Bs is generated as a synthetic panorama image from the synthetic individual camera images EBs1 to EBsn using a stitching algorithm in the data processing unit. In order to generate the synthetic camera image Bs without defects in the stitching algorithm, the virtual individual camera perspectives are selected such that the synthetic individual camera images EBs1 to EBsn overlap. Subsequently, in an eighth process step VS8, visualized in Fig. 9, the synthetic camera image Bs is compared with the original camera image OB. This comparison is performed using pixel registration, whereby at least one pixel BP is identified in the synthetic camera image Bs where the original camera image OB shows the highest degree of similarity with the synthetic camera image Bs. In the comparison, a ninth process step VS9, visualized in Fig. 10, determines a deviation of the original camera image OB from the synthetic camera image Bs. As a measure of the deviation between this image point BP and a corresponding image point of the original camera image OB, a two-dimensional image vector V is determined in the satellite coordinate system within the synthetic camera image B. If the camera alignment ARK and the camera position POSK are correctly calibrated, the content of the synthetic camera image Bs correlates with the original camera image OB. If the deviation exceeds a defined threshold, a miscalibration of the environmental sensing camera 1 is assumed. The image vector V is then transformed from the satellite coordinate system into the vehicle coordinate system FK, forming a three-dimensional vector that describes an updated extrinsic translation and orientation—that is, the deviation of the camera orientation ARK from the virtual camera orientation ARv and the deviation of the camera position POSK from the virtual camera position POSv. The back-transformed vector is then used to update the extrinsic parameters of the environmental sensing camera 1 accordingly. In one possible embodiment of the procedure, it is further provided that if the degree of deviation is greater than a defined threshold, at least one function of at least one driver assistance system of the vehicle 2, which is executed depending on image data captured by the environment detection camera 1, is deactivated. The calibration described above is repeated cyclically and / or interval-based and / or event-based during the operation of vehicle 2 in process step VS10. This online calibration enables the calibration of all environmental sensing cameras 1 of vehicle 2 during its operation and can be performed independently of any scenario preceding vehicle 2.

Claims

Method for calibrating extrinsic parameters of an environment sensing camera (1) of a vehicle (2), wherein an original camera image (OB) of the vehicle's (2) environment is captured by means of the environment sensing camera (1), characterized in that: - a synthetic camera image (Bs) generated from a satellite image (SB) of the vehicle's (2) environment from a virtual camera perspective of the environment sensing camera (1) is requested and received by a central computing unit; - the synthetic camera image (Bs) is compared with the original camera image (OB) and a deviation of the original camera image (OB) from the synthetic camera image (Bs) is determined in the comparison; and - a measure describing the deviation is transformed from the satellite coordinate system into the vehicle coordinate system (FK) and an existing calibration of the extrinsic parameters is checked for correctness on the basis of the transformed measure.wherein a geographic actual vehicle position (POSist) and an actual vehicle orientation (ARist) are determined, and from the actual vehicle position (POSist) by means of displacement to a camera position (POSK) of the environment detection camera (1) and from the actual vehicle orientation (ARist) by means of rotation to a camera orientation (ARK) of the environment detection camera (1) a virtual camera position (POSv) and a virtual camera orientation (ARv) are generated in the vehicle coordinate system (FK),- the virtual camera position (POSv) and the virtual camera orientation (ARv) are transformed into a satellite coordinate system and - the received synthetic camera image (Bs) is generated in the satellite coordinate system according to the virtual camera perspective formed by the virtual camera position (POSv) and the virtual camera orientation (ARv), from which at least one satellite image (SB) of the vehicle's surroundings (2) is requested and received by the central processing unit. The method according to claim 1, characterized in that the comparison is carried out by means of a pixel registration, wherein - in the synthetic camera image (Bs) at least one pixel (BP) is determined in which the original camera image (OB) has the highest agreement with the synthetic camera image (Bs), and - as a measure of the deviation between the pixel (BP) and a corresponding pixel of the original camera image (OB) an image vector (V) in the satellite coordinate system is determined. Method according to one of the preceding claims, characterized in that - in the satellite coordinate system several synthetic single camera images (EBs1 to EBsn) from different virtual single-image camera perspectives are requested and received from the at least one satellite image (SB) and - the synthetic camera image (Bs) is generated from the synthetic single camera images (EBs1 to EBsn) by means of a stitching algorithm. Method according to claim 3, characterized in that the virtual single-image camera perspectives are selected depending on an expected maximum calibration error of the environment detection camera (1). Method according to claim 3 or 4, characterized in that the virtual single-image camera perspectives are selected such that the synthetic single-image camera images (EBs1 to EBsn) overlap. Method according to one of the preceding claims characterized in that in the calibration of the extrinsic parameters at least one deviation of a translation of the environment detection camera (1) from a target translation and one deviation of an orientation of the environment detection camera (1) from a target orientation are determined. Method according to one of the preceding claims, characterized in that, upon detection of an incorrect existing calibration, the existing calibration is replaced by a calibration determined from the transformed measure. Method according to one of the preceding claims, characterized in that, if the transformed dimension falls below a predetermined threshold, the existing calibration is determined to be correct, and if the transformed dimension exceeds a predetermined threshold, the existing calibration is determined to be incorrect. Method according to one of the preceding claims, characterized in that the calibration is carried out cyclically and / or interval-based and / or event-based during driving operation of the vehicle (2) and / or is repeated. Device for calibrating extrinsic parameters of an environment sensing camera (1) of a vehicle (2), wherein the environment sensing camera (1) is configured to capture original camera images (OB) of an environment of the vehicle (2), characterized by: - ​​a data processing unit, which is configured to request and receive from a central computing unit a synthetic camera image (Bs) generated from a satellite image (SB) of an environment of the vehicle (2) from a virtual camera perspective of the environment sensing camera (1),to compare the synthetic camera image (Bs) with the original camera image (OB) and, in the comparison, to determine a deviation of the original camera image (OB) from the synthetic camera image (Bs); to transform a measure describing the deviation from the satellite coordinate system into the vehicle coordinate system (FK) and, using the transformed measure, to check the correctness of an existing calibration of the extrinsic parameters; a position acquisition unit which is designed to determine a geographic actual vehicle position (POSist) and an actual vehicle orientation (ARist).- wherein the data processing unit is configured - to generate a virtual camera position (POSv) and a virtual camera orientation (ARv) in the vehicle coordinate system (FK) from the actual vehicle position (POSist) by means of displacement to a camera position (POSK) of the environment detection camera (1) and from the actual vehicle orientation (ARist) by means of rotation to a camera orientation (ARK) of the environment detection camera (1), - to transform the virtual camera position (POSv) and the virtual camera orientation (ARv) into the satellite coordinate system and - to request and receive from a central processing unit a synthetic camera image (Bs) generated from the satellite image (SB) of the vehicle's environment (2) in the satellite coordinate system according to the virtual camera perspective formed from the virtual camera position (POSv) and the virtual camera orientation (ARv). Method for operating a vehicle (2), wherein - by means of a method according to one of claims 1 to 8, the existing calibration of extrinsic parameters of the environment detection camera (1) of the vehicle (2) is checked for correctness and - if an incorrect existing calibration is detected, at least one function of at least one driver assistance system of the vehicle (2) that is executed depending on image data acquired by means of the environment detection camera (1) is deactivated.

Images