DE102022204313A1 - Method and device for generating an image of the environment for a parking assistant of a vehicle - Google Patents

Abstract

The invention relates to a method for generating an image of the surroundings (21) for a parking assistant of a vehicle (50), wherein the surroundings of the vehicle (50) are recorded at several positions (32) of the vehicle (50) by means of several surrounding cameras (2). Surroundings are detected by means of at least one distance-resolving sensor (3); wherein, based on captured sensor data (11) of the at least one distance-resolving sensor (3), open spaces (40) and raised objects (41) in the environment are identified, a local environment map (30) based on the captured sensor data (11) and/or the identified open areas (40) and the identified raised objects (41), wherein identified open areas (40) and identified raised objects (41) are linked to corresponding camera images (10) depending on the position and the detection direction, an image signal (20) for a Environment image (21) is generated in a top view of the vehicle (50) by projecting the camera images (10) onto the open spaces (40) and objects (41) in the environment map (30) at least partially depending on the position and the detection direction, and whereby this generated image signal (20) is provided. The invention further relates to a device (1) for generating an image (21) of the surroundings.

Description

The invention relates to a method and a device for generating an image of the surroundings for a parking assistant of a vehicle.

In modern vehicles, systems for a camera-based environment view are known, which provide a top view of the vehicle's immediate surroundings. The Area View is intended to make maneuvering easier. For this purpose, the surroundings are recorded by several surrounding cameras and combined to form an image of the surroundings, which is projected onto a planar projection plane. This projection plane corresponds to a standing area of the vehicle. Systems are also known in which the captured camera images are projected onto a bowl shape (“bowl view”) in order to display raised objects further away from the vehicle in a side view. Here, even raised objects are projected onto the flat base of the bowl shape, which also represents the base of the vehicle.

The creation of an image of the environment in a top view is currently not sufficiently solved.

The invention is based on the object of improving a method and a device for generating an image of the surroundings for a parking assistant of a vehicle.

The object is achieved according to the invention by a method with the features of patent claim 1 and a device with the features of patent claim 9. Advantageous embodiments of the invention result from the subclaims.

In particular, a method for generating an image of the environment for a parking assistant of a vehicle is provided, wherein an environment of the vehicle is recorded at several positions of the vehicle using several environment cameras and the environment is recorded using at least one distance-resolving sensor; wherein open spaces and raised objects in the environment are identified based on detected sensor data from the at least one distance-resolving sensor, a local environment map being generated based on the detected sensor data and/or the identified open spaces and the identified raised objects, wherein identified open spaces and identified raised objects are position-dependent and linked to corresponding camera images depending on the detection direction, an image signal for an image of the surroundings in a top view of the vehicle being generated by projecting the camera images onto the open spaces and objects in the surroundings map at least partially depending on the position and depending on the detection direction, and the generated image signal being provided .

Furthermore, in particular, a device for generating an image of the surroundings for a parking assistant of a vehicle is created, comprising a plurality of surrounding cameras that are set up to capture the surroundings of the vehicle, at least one distance-resolving sensor that is set up to capture the surroundings, and a data processing device, wherein the data processing device is set up to identify open spaces and raised objects in the environment based on sensor data recorded at several positions of the vehicle from the at least one distance-resolving sensor, to generate a local map of the surrounding area based on the recorded sensor data and / or the identified open spaces and the identified raised objects To link open spaces and identified raised objects with camera images corresponding to them, depending on the position and the direction of detection, to generate an image signal for an image of the surroundings in a top view of the vehicle, and to project the camera images for this purpose onto the open spaces and objects in the surroundings map at least partially depending on the position and depending on the direction of detection, and that to provide generated image signal.

The method and the device make it possible to generate an improved camera-based image of the environment in a top view. In particular, several positions of the vehicle are taken into account, so that not only an image of the environment for a current position of the vehicle can be provided, but also an image of the environment that goes beyond this. In particular, the multiple positions at which the sensor data and the camera images are captured correspond to positions within a parking scenario. In particular, a history of the environment can be taken into account, that is, a camera-based image of the environment, especially in a parking scenario, can be built up step by step. For this purpose, for example, the vehicle drives (position by position) through the parking scenario (e.g. parking lot or parking garage), with the image of the environment being generated step by step while driving through the parking scenario. The positions are determined, for example, using the vehicle's own odometry and/or retrieved from the odometry. At the several positions, the coordinates of which are determined by the odometry and/or retrieved by it, the surroundings of the vehicle are recorded by means of several surrounding cameras and by means of at least one distance-resolving sensor. This is done in particular at each of the positions. The number and orientation of the surrounding cameras are selected and arranged in such a way that these can capture a complete image of the environment, in particular all horizontal directions around the vehicle, that is, in particular a horizontal angle of 0 to 360°. Based on the recorded sensor data from the at least one distance-resolving sensor, open spaces and raised objects in the surrounding area are identified. A local environment map is generated based on the recorded sensor data and/or from the identified open spaces and the identified raised objects. This is done in particular using methods known per se. The identified open spaces and the identified raised objects are linked to corresponding camera images depending on the position and detection direction. For example, image elements of the camera images captured at the multiple positions can each be linked to corresponding solid angles of the detection areas of the surrounding cameras and stored at the positions. For each of the positions, this enables the image elements in the captured camera images to be assigned to areas in the environment, whereby the camera images can be projected onto the respective areas. An image signal for an image of the surroundings in a top view of the vehicle is generated by projecting the camera images onto the open spaces and objects in the surroundings map at least partially in a position-dependent and detection direction-dependent manner. The generated image signal is provided, for example as an analog or digital signal or digital data packet. The generated image signal can then be displayed, for example, on a display device of the vehicle in order to support a driver of the vehicle when parking and/or maneuvering.

It can be provided that the environment image only takes into account a limited area of the environment, that is, that a history only extends back to a specific point in time or to a specific position in the past. The environment map can, for example, be designed in the manner of a shift register, in which sensor data and camera images recorded up to a specific point in time in the past or up to a specific past position of the vehicle are taken into account. However, it can also be provided to generate the environment image without limitation until a memory space in the data processing device reserved for this purpose is completely occupied.

A distance-resolving sensor is in particular an ultrasonic sensor, a radar sensor or a lidar sensor. In principle, several (different) sensors can also be used. A number and/or an arrangement and/or an orientation of the at least one distance-resolving sensor is selected in particular such that a horizon angle (e.g. azimuth) of 0 to 360° can be detected. A vertical angle range (e.g. elevation) lies, for example, in a range from 0 to approximately 30-45°, but can in principle also cover a different range.

Open spaces are, for example, areas in the environment for which the distance-resolved sensor data show that no object and/or no reflection above a predetermined vertical angle is detected within a predetermined minimum distance from the sensor. The specified minimum distance is, for example, a few meters (e.g. 10 m, 15 m, etc.) and the specified vertical angle is, for example, 5°, 10° or 15° etc.

The local environment map is in particular a three-dimensional environment map, which is described, for example, using a local Cartesian coordinate system.

Projecting is done in particular by placing the camera images as textures over the open spaces and raised objects. The projection is carried out in a perspectively correct form, in particular in a position-dependent and detection direction-dependent manner, so that the image of the environment is generated in a top view in such a way that it ideally corresponds to an image of the environment captured from a bird's eye view using a camera. When projecting, methods known per se, such as stitching and/or brightness and/or contrast adjustments, are used in particular to connect different camera images with one another.

Parts of the device, in particular the data processing device, can be designed individually or collectively as a combination of hardware and software, for example as program code that is executed on a microcontroller or microprocessor. However, it can also be provided that parts are designed individually or combined as an application-specific integrated circuit (ASIC) and/or field-programmable gate array (FPGA).

In one embodiment it is provided that when the environment map is generated from open spaces, a position-dependent ground projection plane is gradually defined for the multiple positions, onto which the camera images linked to the open spaces are projected. Position-dependent means in particular that the ground projection plane is only defined where open spaces have been identified. This allows a floor level for the vehicle's standing area to be gradually defined. This is based on the idea that open spaces cannot be viewed from every position of the vehicle appear as open spaces, for example because an object obscures the open space when viewed from some positions of the vehicle but not from others. However, if this open space can be detected and identified as an open space from several positions or from at least one position, this is sufficient to take this open space into account as such in the environmental image. In particular, once identified as open spaces, areas in the surrounding area are always identified as open spaces, even if they are not identified as open spaces from other positions. The ground projection plane is defined in particular in a local (for example Cartesian) coordinate system of the environment map.

In one embodiment, it is provided that when the environment map is generated from raised objects, at least one position-dependent projection plane is gradually defined for the multiple positions, onto which the camera images associated with the raised objects are projected. This allows raised objects to be displayed as such in the image of the environment in a top view. For example, if there is another vehicle in the vicinity of the vehicle, this other vehicle is identified as a raised object and a projection plane arranged at this highest position is defined for the other vehicle, for example starting from the highest position of the associated sensor data. This projection plane then ideally coincides with a height of a hood or a roof or a trunk lid (in the case of a hatchback) of the other vehicle. The captured camera images are then projected onto the projection planes defined for this purpose, depending on the position and direction of capture, so that a more realistic representation results in the generated image signal or in the image of the environment in a top view.

In one embodiment it is provided that when the at least one projection plane is generated, at least one area is added if the sensor data of the at least one distance-resolving sensor recorded for this area corresponds to another vehicle. This means that hidden areas that cannot be captured by the surrounding cameras can also be represented as such supplemented areas in the surrounding image generated. The idea behind this is that a projection plane for a roof of another vehicle can be added, for example, if the height of a roof edge is known in the surrounding map. The roof is then added at the height that coincides with the edge of the roof. Based on the captured sensor data of the at least one distance-resolving sensor and/or the captured camera images, other vehicles in particular are recognized for this purpose. This is done using methods known per se, such as pattern recognition and/or machine learning. In principle, other objects can also be recognized and supplemented accordingly.

In one embodiment it is provided that a degree of a respective radial distortion of the surrounding cameras is taken into account when detecting and/or projecting. This means that in particular those areas in the captured camera images can be used for projection that show the surroundings, i.e. open spaces and raised objects, undistorted in a high resolution. Since environmental cameras generally have a large detection range (e.g. 180°), imaging optics are used to image this large detection range on an image sensor. The imaging optics image particularly edge areas in a highly distorted manner (“fisheye optics”), so that, compared to the center of the image, a larger solid angle area is imaged on an equal number of sensor elements of the image sensor of the surrounding cameras. This results in increasing distortion and decreasing resolution radially from the center of the image. For example, it can be provided that only areas of the captured camera images are used in which the degree of edge distortion is below a predetermined threshold value (for example specified as the number of image elements per solid angle or as a change in the number of image elements per solid angle, etc.). If only these areas are taken into account when detecting and/or projecting, the overall resolution of the image of the surroundings can be increased. Furthermore, distortion and/or falsification during projection can be reduced or even prevented. For example, if an open area is captured at several positions from different directions, those camera images or areas of these camera images can be used that have the highest resolution or at least a resolution above a predetermined threshold value. Provision can also be made to dynamically adapt such a threshold value depending on the availability of captured camera images. This can be done in particular depending on the resolution with which camera images for identified open spaces and/or identified raised objects were captured and are available. If camera images with higher resolution are available, only these or areas thereof with a resolution above the threshold value and/or below a predetermined degree of edge distortion are used; However, for identified open spaces and/or identified raised objects, only camera images with lower resolution are available (e.g. because they consist of edge-distorted images areas), these are used. The aim here is in particular to always use the highest resolution available. The threshold value can be adjusted dynamically accordingly.

In one embodiment it is provided that only a predetermined inner image area of the captured camera images is used during projection. This also allows radial distortion to be taken into account. In particular, this enables a variant that is particularly easy to implement and requires little effort, since the inner image area can be defined once and then only this inner image area is taken into account when projecting. The inner image area can be defined, for example, by specifying a minimum resolution.

In one embodiment, it is provided that the image signal for areas of the environment image for which no camera data and/or no sensor data is available due to occlusion is extrapolated and/or estimated based on neighboring areas that border on these areas. This means that an image signal can also be provided for these areas that are not captured by the surrounding cameras. The idea behind this is that, for example, a texture for a roof of another vehicle (or another object) can be supplemented and/or appreciated when a texture of a hood, a tailgate and/or another part of the other vehicle (or the other object) is known. The roof is then supplemented with the help of the extrapolated and/or estimated texture or an extrapolated and/or estimated camera image is projected onto the associated projection plane. Based on the captured sensor data of the at least one distance-resolving sensor and/or the captured camera images, in particular other vehicles and/or other objects in the environment are recognized. This is done using methods known per se, such as pattern recognition and/or machine learning. In principle, other objects can also be recognized and supplemented accordingly. It can also be provided that hidden areas are determined based on the recorded sensor data and optical properties of the surrounding cameras.

It can also be provided that identified raised objects, in particular other vehicles, are at least partially rendered in the environment, this being done based on the recorded sensor data of the distance-resolving sensor. A corresponding image signal is generated for the at least partially rendered objects. The rendering is carried out in particular using the data processing device and methods known per se.

In one embodiment, it is provided that, based on the camera images captured at the multiple positions of the vehicle, a three-dimensional point cloud of the surroundings is additionally generated and taken into account when identifying the open spaces and/or the raised objects in the surroundings. This means that distance-resolving data or three-dimensional data of the environment can also be generated and taken into account based on the captured camera images. For example, a monocular Simultaneous Localization and Mapping (SLAM) method (e.g. Direct Sparse Odometry, DSO) or another suitable method can be used.

Further features for the design of the device result from the description of embodiments of the method. The advantages of the device are the same as in the embodiments of the method.

Furthermore, a vehicle is also created in particular, comprising at least one device according to one of the described embodiments. A vehicle is in particular a motor vehicle.

• 1 eine schematische Darstellung einer Ausführungsform der Vorrichtung zum Erzeugen eines Umfeldabbildes für einen Parkassistenten eines Fahrzeugs;
• 2 eine schematische Darstellung zur Verdeutlichung von Sensordaten von entfernungsauflösenden Sensoren;
• 3 eine schematische Darstellung zur Verdeutlichung einer Bodenprojektionsebene und von Projektionsebenen für erhabene Objekte;
• 4 eine schematische Darstellung eines Umfeldabbildes, das dem erzeugten Bildsignal entspricht;
• 5a, 5b schematische Darstellungen zur Verdeutlichung einer Ausführungsform der Vorrichtung und des Verfahrens.

The invention is explained in more detail below using preferred exemplary embodiments with reference to the figures. Show here:

• 1 a schematic representation of an embodiment of the device for generating an image of the surroundings for a parking assistant of a vehicle;
• 2 a schematic representation to illustrate sensor data from distance-resolving sensors;
• 3 a schematic representation to illustrate a floor projection plane and projection planes for raised objects;
• 4 a schematic representation of an environment image that corresponds to the generated image signal;
• 5a , 5b schematic representations to illustrate an embodiment of the device and the method.

The 1 shows a schematic representation of an embodiment of the device 1 for generating an environment image 21 for a parking assistant of a vehicle 50. The device 1 is arranged, for example, in a vehicle 50, in particular a motor vehicle. The device 1 carries out the method described in this disclosure.

The device 1 includes several surrounding cameras 2, at least one distance-resolving sensor 3 and a data processing device 4.

In particular, the device 1 includes four surroundings cameras 2 (for the sake of clarity, there is only one in the 1 shown), which are each designed as 180° capturing cameras with fisheye optics and are arranged on the vehicle 50 in such a way that a horizontal angular range of an environment of 0 to 360° around the vehicle 50 can be captured without gaps.

The at least one distance-resolving sensor 3 includes, for example, an ultrasonic sensor, a radar sensor and/or a lidar sensor. A number and an arrangement of the distance-resolving sensors 3 are selected in particular such that a horizontal angular range of an environment from 0 to 360 ° around the vehicle 50 can be detected without gaps. In particular, the surrounding cameras 2 and the distance-resolving sensors 3 are calibrated to one another, that is, a relative pose (relative position and relative orientation) to one another is known. This enables an assignment of captured sensor data 11 to captured camera images 10.

The data processing device 4 includes a computing device 4-1 and a memory 4-2. The data processing device 4 is set up to identify open spaces and raised objects in the environment based on sensor data 11 of the at least one distance-resolving sensor 3 recorded at several positions of the vehicle 50, a local environment map 30 based on the recorded sensor data 11 and/or the identified open spaces and to generate the identified raised objects and to link identified open spaces and identified raised objects in a position-dependent and detection direction-dependent manner with camera images 10 corresponding thereto. Furthermore, the data processing device 4 is set up to generate an image signal 20 for an image of the surroundings 21 in a top view of the vehicle 50, and the camera images 10 for this at least partially depending on the position and the direction of detection towards the open spaces and objects in the surroundings map 30 project, and provide the generated image signal 20. The provided image signal 20 can, for example, be supplied to a display device 51 of the vehicle 50 and displayed thereon in order to support a driver when parking and/or maneuvering.

The 2 shows a schematic representation to illustrate sensor data from distance-resolving sensors. Shown is an environment map 30 in which three-dimensionally located data points 31-x (only some are marked with a reference number) of an ultrasonic sensor (31-1), a radar sensor (31-2) and a lidar sensor (31-3) are stored. The data points 31-x were recorded by the sensors at several positions 32 and stored in the environment map 30. The data points 31-x mark areas where raised objects 41 are arranged in the environment. There are open spaces 40 in the surrounding area between the raised objects 41. The raised objects 41 are identified, for example, by clustering or combining locally clustered data points 31-x. An accumulation is defined here, for example, as a predetermined density of data points 31-x per solid angle or horizontal angle or area in the environment or as a predetermined signal strength of a reflection of the signal emitted by the sensor, which must be achieved so that a raised object 41 is identified. In the area of open spaces 40, however, such accumulations and/or reflections of a signal emitted by the sensor are missing.

Camera images are captured for all directions around the vehicle 50 at each of the multiple positions 32. Image information in the captured camera images is linked to the data points 31-x depending on the position and direction of capture. In particular, based on this link, in order to generate the environment image 21 in a top view, the camera images are (at least partially) projected onto the open spaces 40 and the raised objects 41 in the environment map 30 depending on the position and the detection direction. In other words, the captured camera images are (at least partially) placed as textures over the open spaces 40 and the raised objects 41 in a position-dependent and capture direction-dependent manner, thereby generating an environment image 21 in a top view.

In particular, it is provided that when generating the environment map 30 from open spaces 40, a position-dependent floor projection plane 33 is gradually defined for the multiple positions 32, onto which the camera images linked to the open spaces 40 are projected. This is schematic in the 3 shown which fundamentally the 2 corresponds, so that the same reference numbers designate the same features and terms.

It is furthermore provided in particular that when generating the environment map 30 from raised objects 41, at least one position-dependent projection plane 34 is defined step by step for the multiple positions 32, onto which the respective projection plane 34 is defined Camera images linked to raised objects 41 are projected. The main thing here is to show the raised objects 41 in the surrounding image 21 ( 1 ) in a top view at a different height (direction perpendicular to the base on which the vehicle 50 stands) than the floor projection plane 34. In particular, more than one projection plane 34 can be provided in order to provide intermediate stages, for example of other vehicles 51 (e.g. intermediate stages for rear, roof, hood, loading area etc.).

It can be provided that when the at least one projection plane 34 is generated, at least one area is added if the sensor data recorded for this area of the at least one distance-resolving sensor corresponds to another vehicle 51. For this purpose, for example, pattern recognition and/or object recognition can be carried out on the recorded sensor data, in particular on the data points 31-x. If it is recognized based on pattern recognition and/or object recognition that a raised object 41 is another vehicle 51, this prior knowledge can be used to identify more distant areas of the other vehicles 51 for which a density of data points 31-x decreases supplement, for example by adding data points 31-x to reproduce an outer contour of the other vehicle 51.

It can be provided that a degree of a respective radial distortion of the surrounding cameras 2 is taken into account when detecting and/or projecting. To determine the degree of radial distortion, empirical tests can be carried out and/or optical property parameters 15 ( 1 ) an imaging optics (“fisheye optics”) are read into the data processing device 4.

It can be provided that only a predetermined inner image area of the captured camera images 10 is used during projection. A suitable inner image area can be determined, for example, based on empirical tests and/or simulations.

It can be provided that the image signal 20 is extrapolated and/or estimated for areas of the environment image 21 for which there is no camera data 10 and/or no sensor data 11 due to occlusion, starting from neighboring areas that border on these areas. For example, the image signal 20 can be used for the side 35 facing away from the vehicle 50 ( 3 ) of the raised objects 41 or other vehicles 51 are supplemented accordingly by image data from camera images 10 that were recorded for the side 36 facing the vehicle 50. For example, a paint color or a texture of the facing sides 36 in captured camera images 10 can be used to extrapolate and/or estimate a texture for the facing side 35.

It can be provided that, based on the camera images 10 captured at the multiple positions 32 of the vehicle 50, a three-dimensional point cloud of the surroundings is additionally generated and taken into account when identifying the open spaces 40 and/or the raised objects 41 in the surroundings. This is done, for example, by means of SLAM methods known per se, for example by means of the Direct Sparse Odometry (DSO) method.

The 4 shows a schematic representation of an environment image 21, which corresponds to the generated image signal 20 (the 4 is merely an example and does not correspond in content to those shown previously 2 and 3 ). Through the device 1 ( 1 ) and the method can provide an improved environment image 21.

That in the 4 The environmental image 21 shown or the image signal 20 was generated in particular with an embodiment of the device 1 and the method, in which it is provided that only a predetermined inner image area of the captured camera images 10 is used during projection. A direct comparison between an embodiment in which outer regions of the camera images 10 were also used, that is, also radially distorted regions due to fisheye optics, and that in the 4 The environment image 21 shown is in the 5a and 5b shown. In the 5a It can be clearly seen that due to the use of the radially distorted outer image areas of the camera images 10, other vehicles 51 appear strongly distorted in the image of the surroundings 21. For the other vehicles 51 that are in the 5b However, this is not the case, since only inner areas of the camera images 10 that are not or only slightly distorted radially were taken into account (ie a viewing area or detection area of the surrounding cameras is particularly laterally restricted).

Reference symbol list

1

contraption

2

Surrounding camera

3

distance-resolving sensor

4

Data processing device

4-1

Computing facility

4-2

Storage

10

Camera image

11

Sensor data

15

optical property parameters

20

Image signal

21

Environment image

30

local environment map

31-1

Data point (ultrasonic sensor)

31-2

Data point (radar sensor)

31-3

Data point (lidar sensor)

32

position

33

Floor projection plane

34

projection plane

35

opposite side

36

facing side

40

Open space

41

sublime object

50

vehicle

51

Display device

Claims (10)

Method for generating an image of the environment (21) for a parking assistant of a vehicle (50), wherein at several positions (32) of the vehicle (50): - an environment of the vehicle (50) is recorded by means of several surrounding cameras (2), - the environment is detected by means of at least one distance-resolving sensor (3), with open spaces (40) and raised objects (41) being identified in the environment based on detected sensor data (11) of the at least one distance-resolving sensor (3), wherein a local environment map (30) is generated based on the recorded sensor data (11) and/or the identified open spaces (40) and the identified raised objects (41), wherein identified open spaces (40) and identified raised objects (41) are linked to corresponding camera images (10) depending on the position and detection direction, an image signal (20) for an image of the surroundings (21) in a top view of the vehicle (50) being generated by the camera images (10) are projected onto the open spaces (40) and objects (41) in the environment map (30) at least partially depending on the position and the direction of detection, and wherein the generated image signal (20) is provided. Procedure according to Claim 1 , characterized in that when the environment map (30) is generated from open spaces, a position-dependent ground projection plane (33) is gradually defined for the several positions (32), onto which the camera images (10) linked to the open spaces (40) are projected. Procedure according to Claim 1 or 2 , characterized in that when the environment map (30) is generated from raised objects (41), at least one position-dependent projection plane (34) is gradually defined for the several positions (32), onto which the camera images linked to the raised objects (41) ( 10) can be projected. Method according to one of the preceding claims, characterized in that when the at least one projection plane (34) is generated, at least one area is supplemented if the sensor data (11) recorded for this area of the at least one distance-resolving sensor (3) is connected to another vehicle (51 ) correspond. Method according to one of the preceding claims, characterized in that a degree of a respective radial distortion of the surrounding cameras (2) is taken into account when detecting and/or projecting. Method according to one of the preceding claims, characterized in that only a predetermined inner image area of the captured camera images (10) is used during projection. Method according to one of the preceding claims, characterized in that the image signal (20) for areas of the environment image (21) for which there is no camera data (10) and/or no sensor data (11) due to occlusion, starting from neighboring areas adjacent to these areas can be extrapolated and/or estimated. Method according to one of the preceding claims, characterized in that , starting from the camera images (10) captured at the several positions (32) of the vehicle (50), a three-dimensional point cloud of the environment is additionally generated and when identifying the open spaces (40) and / or the raised objects (41) in the environment are taken into account. Device (1) for generating an image of the surroundings (21) for a parking assistant of a vehicle (50), comprising: several surroundings cameras (2), set up for Detecting an environment of the vehicle (50), at least one distance-resolving sensor (3) set up to detect the environment, and a data processing device (4), the data processing device (4) being set up to do so, starting from several positions (32) of the vehicle (50) recorded sensor data (11) of the at least one distance-resolving sensor (3) to identify open spaces (40) and raised objects (41) in the environment, a local environment map (30) based on the recorded sensor data (11) and/or the identified ones To generate open spaces (40) and the identified raised objects (41), to link identified open spaces (40) and identified raised objects (41) with camera images (10) corresponding to them in a position-dependent and detection direction-dependent manner, an image signal (20) for an image of the environment (21 ) in a plan view of the vehicle (50), and to project the camera images (10) onto the open spaces (40) and objects (41) in the environment map (30) at least partially depending on the position and detection direction, and the generated image signal (20 ) to provide. Vehicle (50), comprising at least one device (1). Claim 9 .

Images