EP4208372A1

EP4208372A1 - Method for displaying the surroundings of a vehicle on a display device, processing unit, and vehicle

Info

Publication number: EP4208372A1
Application number: EP21769721.8A
Authority: EP
Inventors: Tobias KLINGER; Ralph-Carsten Lülfing
Original assignee: ZF CV Systems Global GmbH
Current assignee: ZF CV Systems Global GmbH
Priority date: 2020-09-02
Filing date: 2021-08-31
Publication date: 2023-07-12
Also published as: WO2022049040A1; CN115968485A; DE102020122908A1; US20230226977A1

Abstract

The invention relates to a method for displaying the surroundings of the vehicle (1) on a display device (7), having at least the following steps: - capturing the surroundings using at least two cameras, each of which has a different field of view, the fields of view of adjacent cameras overlapping; - generating a 360° image (RB) from at least two individual images of different cameras, said individual images being projected onto a reference plane in order to generate the 360° image (RB); - ascertaining depth information on at least one object (O) in the surroundings from at least two different individual images of the same camera by means of triangulation; - generating at least one overlay structure (20) on the basis of the ascertained depth information, wherein each overlay structure (20) is uniquely assigned to an imaged object (O); and - displaying the generated 360° image (RB) containing the at least one object (O) and the at least one generated overlay structure (20) on the display device (7) such that the at least one overlay structure (20) is displayed on and/or adjacently to the respective object (O).

Description

Method for displaying surroundings of a vehicle on a display device, processing unit and vehicle

The invention relates to a method for displaying surroundings of a vehicle, in particular a commercial vehicle, on a display device, a processing unit for carrying out the method, and a vehicle.

It is known from the prior art to create all-round view images from individual images from a number of cameras and to display them to an occupant on a display device. For this purpose, it is known to project the individual images onto a reference plane, for example a horizontal plane under the vehicle, to rotate them accordingly and to create a combined all-round view image from them. The disadvantage of this method is that raised objects that protrude from the reference plane at a certain height are shown tilted backwards in the all-round view image and their proportions are thus distorted. As a result, these objects cannot be intuitively interpreted geometrically by the viewer. This makes orientation based on the all-round view image more difficult.

This effect can be minimized by using additional sensors to determine the height of the detected objects and by using this additional height information to compensate for the distortion in the proportions. The disadvantage here is that such additional sensors are not present in all vehicles and/or additional sensors are expensive. In addition, the computational effort is increased.

For example, DE 10 2017 108 254 B4 describes how to create an image composed of individual images and display it on a display. The individual images are each recorded by several cameras. Depth information relating to the object, in particular a position of the object, can be determined by triangulating two or more images. The objects can also be tracked over time.

According to DE 102015 105 248 A1, provision is made for arranging a first camera on a first part of a vehicle and a second camera on a second part of a vehicle on a two-part vehicle. A first image from the first camera and a second image from the second camera are projected via a homographic matrix onto a ground plane and reference plane, respectively, before the first image and the second image are rotated to create a combined image of the environment.

DE 100 35 223 A1 describes creating an overall image or all-round view image from a number of individual images and projecting the vehicle itself into the all-round view image as an artificial graphic object or overlay structure.

EP 3 293 700 B1 describes reconstructing parts of the environment from a number of individual images from a camera using a structure-from-motion method and thus determining depth information on individual objects. A quality metric is determined in order to obtain an optimal reconstruction of the environment. By combining several cameras or individual images, this can also be done in an all-round view.

DE 102018 100 909 A1 describes how the structure-from-motion method is used to obtain a reconstruction of the current environment and to classify or categorize objects in a neural network.

Proceeding from this, the object of the invention is to specify a method for displaying the surroundings of a vehicle on a display device which can be carried out with little hardware and computing effort and enables an observer to easily orientate himself in the environment. The task is also to specify a processing unit and a vehicle.

This object is achieved by a method according to claim 1 and a processing unit and a vehicle according to the further independent claims. The dependent claims indicate preferred developments.

According to the invention, a method for displaying the surroundings of a vehicle on a display device and also a processing unit for carrying out the method with at least the following steps are provided:

First of all, the environment around the vehicle is recorded, particularly in a close-up area, with at least two cameras, each camera having a different field of view, with the fields of view of neighboring cameras overlapping at least in certain areas, in particular at the edges. An all-round view image is then created from the individual images determined in this way, with each individual image being recorded by a different camera at approximately the same time and the individual images for creating the all-round view image being projected into a reference plane, for example using a homography matrix.

Accordingly, no three-dimensional all-round view image is created from the individual images, but rather an all-round view image in which the surroundings are represented as a two-dimensional projection by appropriately combining the individual images. With a corresponding arrangement and orientation of the cameras, this enables a view of the surroundings from a bird's-eye view when the all-round view image is displayed on the display device.

In a further step of the method according to the invention, depth information is determined for at least one object in the recorded environment, the depth information being obtained by triangulation from at least two different individual images from the same camera, preferably by a so-called structure-from-motion method, the at least one object being imaged in the at least two different individual images, preferably from at least two different points of view. Accordingly, a stereo camera system is not used to determine the depth information. Rather, a perspective reconstruction of the surroundings or the object is obtained merely by image processing of the individual images of a camera, from which the depth information can be derived.

At least one overlay structure is generated as a function of the depth information previously determined by the SfM method, with each overlay structure being uniquely assigned to an imaged object. In a further step, the all-round view image created containing the at least one object and the at least one overlay structure generated is then displayed on the display device in such a way that the at least one overlay structure is displayed on and/or adjacent to the respectively assigned object .

An overlay structure can therefore advantageously be displayed on the display device without additional sensors, based solely on methods of image processing, in particular structure-from-motion. With recourse to the previously determined depth information, this can be displayed in a targeted manner at the position or adjacent to the position on the display device on which the object is also displayed. Even if the respective object in the all-round view image is not displayed on the display device in a geometrically interpretable manner, the overlay structure can help to ensure that the all-round view image can be used for reliable orientation, since the overlay structure contains the essential information relating to an object emphasized on the display device. A distortion of raised objects in the all-round view image that occurs as a result of the projection into the reference plane can at least be compensated for in this way. In addition, the display of the overlay structure can also solve the problem that objects in the overlapping areas of individual images from two adjacent cameras often "disappear" or can be perceived insufficiently, since these objects are tilted backwards in the respective camera perspective and in the all-round view -Image usually does not appear at all or at least cannot be perceived adequately. However, such an object is contained in the overlay structures according to the method described above and is therefore also displayed in the all-round view image.

Provision can preferably be made for a bar and/or a polygon and/or text to be displayed on and/or adjacent to the respectively associated object on the display device as an overlay structure. Accordingly, simple structures can be displayed as an overlay, which are sufficient for highlighting objects for orientation. For example, only one bar can be provided, which is displayed on and/or adjacent to an outer edge of the respectively assigned object on the display device, the bar preferably being perpendicular to an object normal of the respectively assigned object, the object normal being the depth information can be obtained. The outer edge is preferably understood to mean an outer boundary of the respective object, with the bar or the respective overlay structure being superimposed at least on that outer edge or boundary that is closest to the host vehicle. As a result, a boundary of the object in the direction of one's own vehicle can be identified on the all-round view by means of a bar. As a result, the observer can clearly see up to which point in space one's own vehicle can be maneuvered or positioned without touching the object, for example during a parking or maneuvering operation. In addition or as an alternative to the bar, it is preferably provided that a polygon is displayed on the display device as an overlay structure in such a way that the polygon at least partially, preferably completely, spans the respectively assigned object. As a result, the viewer of the display device can not only see the boundary of the respective object in the direction of his own vehicle, but also the extent of the object can be identified, so that the object is highlighted even more clearly on the display device. If an object contour and/or an object shape of the respective object is known from the reconstruction of the environment using the SfM method, the polygon can also be adapted to this object contour or object shape. If the object contour or the object shape is not known, a rectangle can be assumed as a polygon, for example, which covers the object points of the respective object depicted in the all-round view image.

Provision is preferably also made for the at least one overlay structure to be displayed on the display device in a predefined color or in a color that is dependent on the determined depth information with regard to the respectively assigned object. As a result, the viewer can also be provided with information regarding the spatial characteristics of the object by means of color coding. Provision can preferably be made for the color of the at least one overlay structure to be dependent on an object distance between the vehicle and the respectively assigned object, the object distance being obtained from the determined depth information with regard to the respectively assigned object. As a result, an overlay structure on the display device that is further away from one's own vehicle can be given a color, for example, which indicates a low level of danger, for example a green color. On the other hand, an overlay structure that is closer to one's own vehicle on the display device can be given a color that indicates a higher risk, for example a red color. Any color gradations are possible depending on the object distance. Provision can preferably also be made for the color and/or the type of overlay structure of the respective object to be dependent on a movement indicator assigned to the object, with the movement indicator indicating whether the respective object can move, e.g. a person or a vehicle, or is permanently stationary, for example a building or a lantern, the movement indicator being obtained from the ascertained depth information with regard to the respectively associated object. As a result, it can additionally be highlighted on the display device which additional danger can emanate from the displayed object due to a potential movement if this object cannot be clearly recognized directly on the all-round view image due to the distortion.

Provision can preferably be made for an object contour and/or an object shape of the respectively associated object to be determined from the depth information and for the object contour and/or the object shape to be determined using a deep learning algorithm by comparison with known object contours and/or object shapes, a motion indicator for the object in question is derived. As a result, there is no need for computationally expensive object tracking and, instead, the comparison of known objects in a database that is stored in the vehicle or that can be accessed from the vehicle via a corresponding data connection can be used.

Alternatively or in addition (for plausibility checking) it can also be provided that object points on the object depicted in the individual images are tracked over time in order to derive the movement indicator for the object in question. Accordingly, by forming difference images, for example, it can be determined how individual image pixels behave over time and a movement of the object can be inferred from this. Provision can preferably also be made for the at least one overlay structure to be displayed opaquely or at least partially transparently on the display device, so that the at least one overlay structure completely or at least partially displays the all-round view image on and/or adjacent to the respectively assigned object related to the transparency covers. As a result, the respective object can be emphasized by the overlay structure, but at the same time the viewer can be given the opportunity to still recognize the object lying behind it, in order to make a plausible assessment of the danger that the object can pose. Advantageously, the transparency of the overlay structure can also be defined in a manner similar to the color coding as a function of the depth information relating to the object. For example, objects with a larger object distance from the vehicle can be displayed with greater transparency than objects with a small object distance from the vehicle, as a result of which more relevant objects are highlighted more strongly.

It is preferably also provided that the display device has display pixels, with all-round view pixels of the all-round view image being displayed on the display pixels of the display device, with an object contained in the all-round view image being displayed on object pixels, with the object pixels being a subset of the display pixels, wherein the overlay structure assigned to the respective object is displayed or superimposed on and/or adjacent to the respective object pixels on the display device. The overlay structures are preferably superimposed by superimposing an overlay image with at least one overlay structure on the all-round view image with the at least one object on the display device in such a way that the overlay structure assigned to the respective object is on and/or adjacent to the respective object pixels is displayed on the display device.

Accordingly, a pre-processing is first carried out using the depth information in such a way that an overlay image is created in which preferably Finally, the overlay structures are displayed at the positions at which or adjacent to which the objects are displayed in the all-round view image. By means of an addition or multiplication or any other operation, these two images can then be displayed simultaneously on the display device in order to achieve the superimposition according to the invention.

Alternatively or additionally, however, it can also be provided that the all-round view image itself contains the at least one overlay structure, with the all-round view image being adapted in this way at and/or adjacent to the all-round view pixels on which an object is depicted that the overlay structure assigned to the respective object is displayed on and/or adjacent to the respective object pixels on the display device. In this case, the display device is only sent an image for display, which is “provided” with the overlay structures in advance, in that the pixels are appropriately “manipulated” in order to display the overlay structures in this image.

Provision is preferably also made for the at least two individual images from which the depth information is determined by triangulation to be recorded by the same camera from at least two different points of view, with the depth information being determined by triangulation as a function of a base length between the at least two points of view. This ensures reliable determination of the depth information within the scope of the SfM method. Provision can preferably be made for the camera to be brought into the different positions by a movement of the vehicle itself, i.e. a change in the driving dynamics of the vehicle, or by an active adjustment of the camera without changing the driving dynamics of the vehicle.

Provision is preferably also made for the surroundings to be depicted within an all-round vision area in the all-round vision image, with the all-round vision region being larger than the vision regions of the individual carcasses. ras, with a viewing angle of the all-round view area being 360°, and with the all-round view image being composed of at least two individual images recorded approximately at the same time by different cameras. This enables a preferably complete imaging of the entire environment around the vehicle at the current point in time and a superimposition of overlay structures for the entire all-round visibility field of view.

Provision is preferably also made for isolines assigned to the vehicle to be displayed on the display device as a further overlay structure, with the isolines being displayed at fixed iso-distances from a vehicle exterior of one's own vehicle. The vehicle itself can thus also be viewed as an object to which overlay structures are assigned, which the viewer can use to orientate himself. The position of the isolines depends on the spatial information that is determined via the individual images. For example, the isolines can be displayed at intervals of 1 m around the host vehicle. Depending on the iso-distance, the overlay structures can also be color-coded with color gradations, so that iso-lines closer to the vehicle are shown in red and iso-lines further away are shown in green, with a corresponding color gradient for iso-lines in between. The transparency of the isolines can also vary depending on the iso distance in order to make the distances to the vehicle easily recognizable to the viewer.

A vehicle according to the invention, in which the method according to the invention can be carried out, therefore has at least two cameras, each camera having a different field of view, with the fields of view of adjacent cameras overlapping at least partially, in particular at the edges. Furthermore, a display device and a processing unit according to the invention are provided in the vehicle, the display device being designed to display a created all-round view image containing at least one object and at least one generated overlay structure—as part of the all-round view image or as a separate overlay image - to represent in such a way that the at least one overlay structure is displayed on and/or adjacent to the respectively assigned object.

It is preferably provided that each individual camera has a field of vision with a viewing angle of greater than or equal to 120°, in particular greater than or equal to 170°, the camera being designed, for example, as a fisheye camera that is selected from the group on at least two sides of the vehicle consisting of a front, a back or at least one long side. With such arrangements, an almost complete recording of the surroundings is possible with a corresponding viewing angle, in order to present them in a bird's-eye view. For this purpose, the individual cameras are preferably aimed at a close-up area of the environment, i.e. at the ground, in order to enable a bird's-eye view.

The invention is explained in more detail below using an exemplary embodiment. Show it:

1 shows a schematic of a vehicle for carrying out the method according to the invention;

2 shows a detailed view of the recorded individual images;

2a shows a detailed view of an object imaged in two individual images from a single camera;

FIG. 2b shows a detailed view of a display device in the vehicle according to FIG. 1 ; and

3 shows a detailed view of the environment displayed on the display device. FIG. 1 shows a vehicle 1, in particular a commercial vehicle, which, according to the embodiment shown, has a front camera 3a on a front side 2a, for example in the roof liner, and a rear camera 3b on a rear side 2b. Furthermore, side cameras 3c are arranged on the longitudinal sides 2c of the vehicle 1, for example on the mirrors. Additional cameras 3 (not shown) can also be provided in the vehicle 1 in order to capture an environment U, in particular a close-up area N (environment up to 10 m away from the vehicle 1).

Each camera 3 has a field of vision 4, with a front field of vision 4a of the front camera 3a looking forward, a rear field of vision 4b of the rear field camera 3b looking backward and a side field of vision 4c of the side cameras 3c to the respective side of the Vehicle 1 are aligned. In order to be able to capture the relevant part of the environment U, in particular the close-up area N, the cameras 3 are aligned towards the ground on which the vehicle 1 is moving.

The number and the position of the cameras 3 is preferably chosen such that the viewing areas 4 of adjacent cameras 3 overlap in the close-up area N, so that all viewing areas 4 together can cover the close-up area N without gaps and thus over the entire area. For this purpose, the cameras 3 can each be designed, for example, as fisheye cameras, which can each cover a field of view 4 with a viewing angle W of equal to or greater than 170°.

Each camera 3 outputs image signals SB that characterize the surroundings U in the field of view 4 that are imaged on the sensor of the respective camera 3 . The image signals SB are output to a processing unit 6, the processing unit 6 being designed to generate 3 individual images EBk (running index k) for each camera on the basis of the image signals SB. According to FIG. 2, the kth individual image EBk has a number Ni of individual image pixels EBkPi (running index i from 0 to Ni) on which the surroundings U are imaged. Certain single image pixels EBkPi are assigned to object points PPn (running index n) which belong to an object 0 in the environment U according to FIG.

In the processing unit 6, the individual images EBk from different cameras 3 are converted by projecting the individual images EBk into a reference plane RE, for example a horizontal plane under the vehicle 1 (cf. a plane parallel to the plane spanned by xO and yO in FIG. 2a), an all-round view image RB with a number Np of all-round view image pixels RBPp (running index p from 0 to Np) is created by a perspective transformation, eg as a function of a homography matrix, using an all-round view algorithm A1. In the all-round view image RB, the surroundings U around the vehicle 1 are displayed without gaps on all sides, at least in the close-up range N (cf. FIG. 3), which corresponds to a viewing angle W of 360°. Therefore, the all-round view image RB results in an all-round view field of view 4R, which is larger than the individual fields of view 4 of the cameras 3. This all-round view image RB of the close-up area N can be shown to an occupant, for example the driver of the vehicle 1, on a display device 7 be issued, so that the driver can orientate himself, for example, when parking or maneuvering. The environment U can be presented to the viewer in a bird's-eye view.

According to Fig. 2b, the display device 7 has a number Nm of display pixels APm (running index m from 0 to Nm), with each all-round view image pixel RBPp being displayed on a specific display pixel APm of the display device 7, so that the viewer All-round view image RB visible on the display device 7 results. A dynamic subset of the display pixels APm are object pixels OAPq (running index q), on which an object O from the environment U is displayed (only schematically in FIG. 2b). All-round view image pixels RBPp, on which a specific object O from the environment U or a specific object point PPn is imaged, are therefore assigned to the object pixels OAPq. With such a display, the application of the all-round view algorithm A1 can result in distortions occurring at the image edges of the all-round view image RB. In order to counteract this, it is provided according to the invention that further information, which results from depth information TI for the objects 0 shown, is superimposed on the created all-round view image RB.

The depth information TI is obtained from a number of individual images EBk from a single camera 3 using the so-called Structure-from-Motion (SfM) method. Depth information TI is therefore extracted camera by camera for each camera 3 individually. In the SfM method, the relevant three-dimensional object 0 in the environment U with its object points PPn is recorded by the respective camera 3 from at least two different viewpoints SP1, SP2, as shown in FIG. 2a. The depth information TI relating to the respective three-dimensional object 0 can then be obtained by triangulation T:

For this purpose, image coordinates xB, yB are determined for at least one first single-image pixel EB1 P1 in a first single image EB1, e.g. of the front camera 3a, and at least one first single-image pixel EB2P1 in a second single image EB2, also of the front camera 3a. Both individual images EB2 are recorded by the front camera 3a at different positions SP1, SP2, ie the vehicle 1 or the front camera 3a moves by a base length L between the individual images EB1, EB2. The first two individual image pixels EB1 P1 , EB2P1 are selected in the respective individual images EB1, EB2 in such a way that they are assigned to the same object point PPn on the three-dimensional object 0 that is depicted in each case.

In this way, one or more pairs of individual image pixels EB1Pi, EB2Pi for one or more object points PPn can be determined for one or more objects 0 in the environment U. To simplify the process, a certain number of frame Pixels EB1 Pi, EB2Pi in the respective individual image EB1, EB2 are combined in a feature point MP1, MP2 (see Fig. 2), the individual image pixels EB1 Pi, EB2Pi to be combined being selected in such a way that the respective feature point MP1, MP2 has a specific clearly localizable feature M is assigned to the three-dimensional object 0. The feature M can be, for example, a corner ME or an outer edge MK on the three-dimensional object 0 (cf. FIG. 2a).

In an approximation, the absolute, actual object coordinates xO, yO, zO (world coordinates) of the three-dimensional Object 0 or the object point PPj or the feature M are calculated or estimated. In order to be able to carry out the triangulation T, a correspondingly determined base length L between the positions SP1, SP2 of the front camera 3a is used.

From the object coordinates xO, yO, zO determined from this, both a position and an orientation, ie a pose, of the vehicle 1 relative to the respective three-dimensional object 0 can then be determined from geometric considerations if the triangulation T for a sufficient number of object points PPi or characteristics M of an object 0 was carried out. Based on this, the processing unit 6 can also at least estimate an object shape FO and/or an object contour CO if the exact object coordinates xO, yO, zO of a plurality of object points PPj or features M of an object 0 are known. The object shape FO and/or the object contour CO can be supplied to a deep learning algorithm A2 for later processing.

In the manner described, objects 0 and their object coordinates xO, yO, zO can also be detected by any other camera 3 in the vehicle 1 and their position and orientation in space can be determined via this. In order to determine the depth information TI even more precisely, it can additionally be provided that more than two individual images EB1, EB2 are recorded with the respective camera 3 and evaluated by triangulation T as described above and/or a bundle adjustment is additionally carried out.

As already described, the object 0 for the SfM method can be viewed from at least two different viewpoints SP1, SP2 by the respective camera 3, as shown schematically in FIG. 2a. For this purpose, the respective camera 3 is to be moved into the different positions SP1, SP2 in a controlled manner. Based on odometry data OD, it can be determined which base length L results between the positions SP1, SP2 from this movement. Different methods can be used for this:

If the entire vehicle 1 is in motion, this already results in a movement of the respective camera 3. This means that the vehicle 1 is actively set in motion in its entirety, for example by the drive system, or passively, for example by a slope will. If at least two individual images EB1, EB2 are recorded by the respective camera 3 during this movement within a time offset, the base length L can be determined using odometry data OD, from which the vehicle movement and thus also the camera movement can be derived. The two positions SP1, SP2 assigned to the individual images EB1, EB2 are thus determined by odometry.

For example, wheel speed signals S13 from active and/or passive wheel speed sensors 13 on the wheels of vehicle 1 can be used as odometry data OD. Depending on the time offset, these can be used to determine how far the vehicle 1 or the respective camera 3 has moved between the positions SP1, SP2, from which the Base length L follows. In order to make the odometric determination of the base length L more precise when the vehicle 1 is moving, further odometry data OD available in the vehicle 1 can be accessed. For example, a steering angle LW and/or a yaw rate G, which are correspondingly determined by sensors or analytically, can be used in order to also take into account the rotational movement of the vehicle 4 .

However, it is not necessary to only use vehicle odometry, i.e. the evaluation of the vehicle movement using motion sensors on vehicle 1. In addition or as an alternative, visual odometry can also be used. In visual odometry, a camera position can be continuously determined from the image signals SB of the respective camera 3 or from information in the captured individual images EB1, EB2, insofar as object coordinates xO, yO, zO of a specific object point PPn, for example, are known at least at the beginning. The odometry data OD can therefore also contain a dependency on the camera position determined in this way, since the vehicle movement between the two positions SP1, SP2 or directly also the base length L can be derived from this.

In principle, however, an active adjustment of the camera 3 can also be provided without changing the state of motion of the entire vehicle 1 . Accordingly, any movements of the respective camera 3 are possible in order to bring them into different positions SP1, SP2 in a controlled and measurable manner.

Depending on the depth information TI, which could be determined by the SfM method for a specific object 0, overlay structures 20 can subsequently be superimposed on the all-round view image RB, as shown in FIG. The superimposition can take place in such a way that the display device 7 is transmitted the all-round view image RB via an all-round view image signal SRB and an overlay image OB to be superimposed with the respective overlay structures 20 is transmitted via an overlay signal SO. The display device Device 7 then displays both images RB, OB on the corresponding display pixels APm, for example by pixel addition or pixel multiplication or any other pixel operation. Alternatively, the all-round view image RB can also be changed or adjusted directly in the processing unit 6 at the corresponding all-round view pixels RBPp, so that the display device 7 can use the all-round view image signal SRB to show an all-round view image RB that contains the overlay structures 20 includes, is transmitted for display.

In this case, the additional overlay structures 20 are uniquely assigned to a specific object O in the environment U. As a result, the observer can be presented with additional information about the respective object O, which makes orientation in the area U based on the display more convenient. For this purpose, the overlay structure 20 can be, for example, a bar 20a and/or a polygon 20b and/or a text 20c, which can additionally be encoded depending on the respectively associated depth information TI.

The superimposition takes place in such a way that the overlay structure 20 appears on or adjacent to the object pixels OAPq of the display device 7 which are assigned to the object O. FIG. The respective object pixels OAPq can be dynamically identified by the processing unit 6 via the all-round view algorithm A1. Based on this, the overlay image OB can be created or the all-round view pixels RBPp of the all-round view image RB can be changed or adjusted directly, so that the respective overlay structure 20 is on or adjacent to the respective object O on the display device 7 appears.

For example, a bar 20a can be displayed on the display pixels APm of the display device 7 that is located on or adjacent to an outer edge MK of the respective object O that is closest to the host vehicle 1 . The alignment of the bar 20a can be chosen such that the bar 20a is perpendicular to an object normal NE, as shown in FIG Delimitation of the object 0 indicates if the object 0 has no straight outer edge MK, for example. The object normal NO can be estimated from the depth information TI on this object, ie from the position and the orientation, which follows from the SfM method.

In order to emphasize the position of a bar 20a assigned to the object 0 on the display device 7, the object pixels OAPq of the object O, to which the bar 20a is also assigned, can be colored in a fixed color F. As a result, the object O itself is displayed more clearly, so that possible distortions in the display of the object O are perceived less. A polygon 20b with the object shape OF or the object contour OC in a specific color F is thus superimposed as a further overlay structure 20 on the all-round view image RB in the area of the object pixels OAPq. If the object shape OF or the object contour OC cannot be clearly determined in the SfM method, only a rectangle can be assumed for the object O, which then runs “behind” the bar 20a as seen from the vehicle 1 . In this case, the further overlay structure 20 is a polygon 20b with four corners.

Black, for example, can be selected as the color F. However, the color F can also be selected as a function of an object distance OA from the respective object O. The bar 20a itself can also be color-coded depending on the object distance OA from the respective object O. The object distance OA between the vehicle 1 and the object O also follows from the depth information TI on this object obtained via the SfM method, i.e. from the position and the orientation.

If an object distance OA of less than 1 m is determined in the SfM method, the color F of the respective overlay structure 20, ie the bar 20a and/or the polygon 20b, can be displayed in a warning color, in particular red, for example will. If the object distance OA of the object O is in a range between 1 m and 5 m, yellow can be used as the color F for this object O associated overlay structure 20 can be used. In the case of object distances OA greater than 5 m, green can be provided as the color F. In this way, the danger emanating from the respective object O can be shown to the observer in a clear manner. Since the depth information TI is obtained from the individual images EBk, the distortions from the all-round view algorithm A1 have no influence on the overlay structures 20 and can therefore be displayed at the correct position on the display device 7 starting from the host vehicle 1 .

Furthermore, the respective overlay structure 20 can be displayed opaquely or at least partially transparently on the display device 7, so that the at least one overlay structure 20 completely or at least partially shows the all-round view image RB on and/or adjacent to the respectively assigned object O in terms of transparency.

In addition, the color F of the overlay structure 20 can be selected as a function of a movement indicator B. As described, an object contour OC and/or an object shape OF for the respectively detected object O can be determined using the SfM method. However, no direct conclusions about the dynamics of the object O can be drawn from the SfM method. However, if the object contour OC and/or the object shape OF is supplied to a deep learning algorithm A2 in the processing unit 6, at least one classification of the object O can take place, from which the possible dynamics of the object O can then be inferred .

The object contour OC and/or the object shape OF of the respective object O can be compared with known objects. These can be stored in a database, which is stored, for example, in a memory fixed to the vehicle or which can be accessed from vehicle 1 via a mobile data connection. Based on the entries of known objects in the database, it can be determined whether the detected object O is a person or a building or a vehicle or the like. Each detected object 0 can be assigned a movement indicator B stored in the database, which indicates whether and how the object 0 normally moves in the environment U. From this it can be concluded whether increased attention to the object 0, eg in the case of people, is required. Correspondingly, the overlay structure 20, for example the bar 20a and/or the polygon 20b or another overlay structure 20, can be coded according to the movement indicator B, for example in color. In addition, a text 20c, for example in the form of an “!” (exclamation mark), etc., can be superimposed as a further overlay structure 20.

However, the movement indicator B of an object O can also be estimated by chronologically tracking individual image pixels EBkPi, which are assigned to an object point PPn in the environment U. This can be done, for example, by forming a difference between successive individual images EBk pixel by pixel. A movement of the respective object O can also be inferred from this.

Furthermore, isolines 20d (see FIG. 3) can be superimposed as overlay structures 20, each of which characterizes a fixed iso-distance A1 to a vehicle exterior 1a.

List of reference symbols (part of the description)

1 vehicle

1a vehicle exterior

2a front

2b back

2c long side

3 camera

3a front camera

3b rear camera

3c side camera

4 viewing area

4a Front viewing area

4b rear view area

4c side viewing area

4R all-round vision field of view

6 processing unit

7 display device

13 wheel speed sensor

20 overlay structure

20a bar

20b polygon

20c text

20d isoline

A1 surround view algorithm

A2 deep learning algorithm

AI Iso distance

APm w. display pixel

B Movement indicator

EBk k. Still image of the camera 3

EBkPi i. Frame pixel of the k. frame EBk

F color G yaw rate

L base length

LW steering angle

M feature

ME corner

MK outer edge

MP1 feature point in the first frame E1

MP2 feature point in second frame E2

N close range

Ni Number of frame pixels EBkPi

Nm Number of display pixels APm

Np number of surround view image pixels RBPp

0 object

OAPq q. object pixel

OA object distance

OB overlay image

OC object contour

OD odometry data

OF object form

ON object normal

PPn n. object point of an object 0

RB all-round view image

RBPp p. Surround view image pixel

RE reference plane

S13 wheel speed signal

SB image signal

SO overlay signal

SP1 , SP2 Camera point of view 3

SRB surround view image signal

TI depth information

T triangulation

U environment xB, yB image coordinates xO, yO, zO object coordinates i, k, m, n, p, q index

Claims

- 25 - Claims

1. A method for displaying an environment (U) of a vehicle (1), in particular a commercial vehicle, on a display device (7), with at least the following steps:

- Recording the surroundings (U) with at least two cameras (3), each camera (3) having a different field of view (4), with fields of view (4) of adjacent cameras (3) overlapping at least in certain areas;

- Creating an all-round view image (RB) from at least two individual images (EBk), each individual image (EBk) being recorded by a different camera (3) and the individual images (EBk) for creating the all-round view image (RB) in a reference plane (RE) to be projected;

- Determining depth information (TI) of at least one object (0) in the recorded environment (U), the depth information (TI) being determined by triangulation (T) from at least two different individual images (EBk) of the same camera (3), wherein the at least one object (O) is imaged in the at least two different individual images (EBk); and

- Generating at least one overlay structure (20) as a function of the depth information (TI) determined, each overlay structure (20) being uniquely associated with an imaged object (O); and

- Displaying the created all-round view image (RB) containing the at least one object (0) and the at least one generated overlay structure (20) on the display device (7) in such a way that the at least one overlay structure (20) is on and/or or is displayed adjacent to the respectively assigned object (0).

2. The method according to claim 1, characterized in that on the display device (7) as an overlay structure (20) a bar (20a) and / or a polygon (20b) and / or text (20c) on and / or adjacent for the assigned object (O) is displayed. Method according to Claim 2, characterized in that the bar (20a) is displayed on and/or adjacent to an outer edge (MK) of the respectively assigned object (0) on the display device (7), the bar (20a) preferably being vertical lies to an object normal (NO) of the respectively associated object (O), the object normal (NO) being obtained from the depth information (TI). Method according to Claim 2 or 3, characterized in that a polygon (20b) is displayed on the display device (7) as an overlay structure (20) in such a way that the polygon (20b) at least partially, preferably completely covered. Method according to one of the preceding claims, characterized in that the at least one overlay structure (20) is displayed on the display device (7) in a predefined color (F) or in a color (F) which is determined by the determined depth information ( TI) is dependent on the respectively assigned object (O). The method according to claim 5, characterized in that the color (F) of the at least one overlay structure (20) depends on an object distance (OA) between the vehicle (1) and the respectively associated object (O), wherein the Object distance (OA) from the determined depth information (TI) with respect to the respectively associated object (0) is obtained. Method according to Claim 5 or 6, characterized in that the color (F) and/or the type of overlay structure (20) for the respective object (0) depends on a movement indicator (B) assigned to the object (0), the movement indicator (B) indicating whether the object (0) can move or is permanently stationary, the movement indicator (B) being obtained from the determined depth information (TI) with regard to the respectively assigned object (0). Method according to Claim 7, characterized in that an object contour (OC) and/or an object shape (OF) of the respectively assigned object (0) is determined from the depth information (TI) and from the object contour (OC) and/or the object shape (OF) via a deep learning algorithm (A2) by comparison with known object contours (OC) and/or object shapes (OF) a movement indicator (B) for the object in question (O ) is derived. Method according to Claim 7 or 8, characterized in that object points (PPn) on the object (O) depicted in the individual images (EBk) are tracked over time in order to derive the movement indicator (B) for the relevant object (O). Method according to one of the preceding claims, characterized in that the at least one overlay structure (20) on the display device (7) is displayed opaque or at least partially transparent, so that the at least one overlay structure (20) the all-round view Image (RB) on and/or adjacent to the respective associated object (O) completely or at least partially covers. Method according to one of the preceding claims, characterized in that the display device (7) has display pixels (APm), all-round view pixels (RBPp) of the all-round view image (RB) being displayed on the display pixels (APm) of the display device (7), wherein an object (O) contained in the all-round view image (RB) is displayed on object pixels (OAPq), the object pixels (AWPq) being a subset of the display pixels (APm), the overlay structure assigned to the respective object (0). (20) is displayed on and/or adjacent to the respective object pixels (OAPq) on the display device (7). - 28 - Method according to claim 11, characterized in that the all-round view image (RB) with the at least one object (0) on the display device (7) has an overlay image (OB) with at least one overlay structure (20 ) is superimposed in such a way that the overlay structure (20) assigned to the respective object (O) is displayed on and/or adjacent to the respective object pixels (OAPq) on the display device (7). Method according to Claim 11 or 12, characterized in that the all-round view image (RB) contains the at least one overlay structure (20), the all-round view image (RB) being at and/or adjacent to the all-round view pixels (RBPp ), on which an object (O) is depicted, is adjusted in such a way that the overlay structure (20) assigned to the respective object (O) is displayed on and/or adjacent to the respective object pixels (OAPq) on the display device (7). will. Method according to one of the preceding claims, characterized in that the at least two individual images (EBk), from which the depth information (TI) is determined by triangulation (T), from at least two different points of view (SP1, SP2) from the same camera (3 ) are recorded, the depth information (TI) being determined by triangulation (T) as a function of a base length (L) between the at least two positions (SP1, SP2). Method according to one of the preceding claims, characterized in that the surroundings (U) are mapped within an all-round vision area (4R) in the all-round vision image (RB), the all-round vision region (4R) being larger than the vision regions (4 ) of the individual cameras (3), with a viewing angle (W) of the all-round view area (4R) being 360°, and with the all-round view image (RB) consisting of at least two individual images (EBk) recorded at approximately the same time by different cameras (3 ) is compiled. - 29 - Method according to one of the preceding claims, characterized in that isolines (20d) assigned to the vehicle (1) are displayed on the display device (7) as an overlay structure (20), the isolines (20d) being in defined iso Distances (AI) to a vehicle outside (1a) of your own vehicle (1) are displayed. Processing unit (6) for carrying out a method according to one of the preceding claims. Vehicle (1) with at least two cameras (3), each camera (3) having a different field of view (4), with fields of view (4) from adjacent cameras (3) overlapping at least in certain areas, a display device (7) and a processing unit (6) according to claim 17, wherein the display device (7) is designed to display a created all-round view image (RB) containing at least one object (O) and at least one generated overlay structure (20) in such a way that the at least one overlay structure (20) is displayed on and/or adjacent to the respectively associated object (O). Vehicle (1) according to Claim 18, characterized in that each individual camera (3) has a field of vision (4) with a viewing angle (W) of greater than or equal to 120°, in particular greater than or equal to 170°.