Method and device for determining the distance of a vehicle to an object by means of a monocular camera

Abstract

The present application relates to the field of scene recognition of a vehicle. The present application provides a method for determining the distance from a vehicle to an object by means of a monocular camera arranged on the vehicle, comprising the following steps: S1: taking a first image of an object in front of the vehicle by means of the monocular camera; S2: changing the distance from the monocular camera to the object along the driving direction of the vehicle, and then taking a second image of the object by means of the monocular camera; S3: performing stereovision analysis on the object based on the first image and the second image, and determining the distance from the vehicle to the object according to the result of the stereovision analysis. The present application also provides an apparatus for determining the distance from a vehicle to an object by means of a monocular camera and a vehicle. By changing the distance from the monocular camera to the object in the driving direction of the vehicle, a difference in viewing angle is provided for the two image detections, so that stereovision analysis can be completed based on a single camera and accurate distance detection can be achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to a method for determining a distance of a vehicle to an object by means of a monocular camera, a device for determining a distance of a vehicle to an object by means of a monocular camera and a vehicle. BACKGROUND

[0002] With the continuous development of vehicle visual perception technology, a stereo camera equipped on a vehicle can generate a spatial image of the surrounding environment of the vehicle and evaluate it in detail. The stereo camera usually comprises a plurality of monocular cameras arranged at a distance from each other in a horizontal plane perpendicular to the optical axis. Thereby, a spatial image of the surrounding environment of the vehicle can be generated by superimposing the images of the plurality of cameras on each other, which enables a very accurate determination of the 3D profile features of a detected object and a reliable determination of the distance of the vehicle relative to the object. This cannot be achieved by a single monocular camera without additional control measures or can only be achieved in a limited manner for rough distance detection.

[0003] However, one of the disadvantages of such a stereo camera is its high cost, and in addition, the plurality of cameras of the stereo camera must be calibrated and coordinated frequently, which is time-consuming and can introduce further errors.

[0004] At present, a method for distance detection by means of a monocular camera is proposed in the prior art, in which, in order to create different perspectives of the object for the monocular camera to take pictures, a lateral shift of the vehicle itself must be made each time the distance detection is performed, which will affect the driving experience of the user in a noticeable manner during normal driving.

[0005] In addition, a method for controlling a vehicle is also known, in which distance jumps occurring when a monocular camera performs a distance measurement are collected, and by analyzing these distance jumps, the length of the trailer part of a vehicle in front can be estimated and used for the adaptive cruise function of the own vehicle. However, since the generation of distance jumps has a certain randomness, the accuracy of the estimated length of the trailer cannot be ensured.

[0006] Therefore, the above-mentioned solutions still have many deficiencies. Against this background, it is desirable to provide an improved distance measurement scheme based on a monocular camera, which aims to ensure the accuracy of the detection result while reducing the interference with the normal driving experience during the implementation of the scheme. SUMMARY

[0007] The present application aims to provide a method for determining a distance of a vehicle to an object by means of a monocular camera, a device for determining a distance of a vehicle to an object by means of a monocular camera and a vehicle, which at least solve some of the problems in the prior art.

[0008] According to a first aspect of the present application, a method for determining a distance from a vehicle to an object by means of a monocular camera arranged on the vehicle is proposed, the method comprising the steps of:

[0009] S1: taking a first image of an object located in front of the vehicle by means of the monocular camera;

[0010] S2: changing the distance from the monocular camera to the object along the driving direction of the vehicle, and then taking a second image of the object by means of the monocular camera; and

[0011] S3: performing a stereo vision analysis on the object based on the first image and the second image, and determining the distance from the vehicle to the object according to the result of the stereo vision analysis.

[0012] The present application particularly comprises the following technical concept: by changing the distance to the object in front of the vehicle along the driving direction of the vehicle, a difference in perspective is provided for the two image taking of the monocular camera, so that a stereo vision analysis can be completed in the vehicle based on only a single camera, thereby saving hardware costs while ensuring high-precision distance detection. In addition, only a longitudinal offset along the driving direction of the vehicle is utilized in performing the distance detection, rather than a lateral offset, so that the passengers in the vehicle do not feel frequent lane changes or significant lateral movements, thereby improving the riding comfort.

[0013] Optionally, the step S3 comprises: reconstructing 3D contour information of the object by means of a Structure from Motion method based on the first image and the second image, and determining the distance from the vehicle to a protrusion on the object based on the 3D contour information.

[0014] Thereby, the following technical advantages are achieved: in the case of combining the Structure from Motion method, the 3D information of the target object is allowed to be calculated from a 2D image sequence from different perspectives. By understanding such 3D information, no matter the appearance of the object in front, the distance to any point or surface on the 3D contour of the target object can be very accurately determined for the vehicle, thereby improving the reliability of distance detection.

[0015] Optionally, the method further comprises the following step before the step S2:

[0016] identifying a height of a protrusion on the object relative to the ground based on the first image;

[0017] comparing the height of the protrusion relative to the ground with a height of the vehicle itself; and

[0018] ignoring the protrusion when determining the distance when the height of the protrusion relative to the ground is greater than the height of the vehicle itself.

[0019] Thereby, the following technical advantage is achieved: By taking into account the ground clearance of the protruding structure of the object in front, a preliminary risk screening of the object can be performed. Thereby, protruding structures that are higher than the vehicle itself and thus less dangerous to the vehicle can be excluded or considered less strongly when determining the distance, thus saving computing effort.

[0020] Optionally, the step S2 comprises keeping the position of the monocular camera relative to the vehicle constant and changing the distance of the vehicle to the object along the driving direction.

[0021] Thereby, the following technical advantage is achieved: There is no need to provide the monocular camera to be movable relative to the vehicle, but only the change of the distance needs to be achieved by appropriate control of the vehicle speed, thus simplifying the installation complexity of the monocular camera. Furthermore, since no lateral displacement of the vehicle is involved, the vehicle does not need to be deviated from the current lane. Thus, a higher comfort is achieved in general, improving the acceptance of the autonomous driving function and / or the driving assistance function of the vehicle.

[0022] Optionally, the step S2 comprises keeping the distance of the vehicle to the object constant and moving the monocular camera in the driving direction of the vehicle to change the distance of the monocular camera to the object.

[0023] Thereby, the following technical advantage is achieved: By controlling the movement of the monocular camera relative to the vehicle without intervention of the vehicle itself, the driving safety during the distance detection is improved and less disturbance to other traffic participants is caused, while the longitudinal acceleration / deceleration of the vehicle is reduced, further improving the ride comfort for the passengers in the vehicle.

[0024] Optionally, the method further comprises, before the step S2, the step of identifying the object in front of the vehicle by means of a trained classifier and / or artificial neural network.

[0025] Thereby, the following technical advantage is achieved: By pre-identifying the type of the object in front, this information can be used to better assist the 3D profile analysis, improving the reliability of the distance detection.

[0026] Optionally, the method further comprises the step of performing a plausibility check of the result of the identification by means of typical identifiers and / or labels on the object.

[0027] Thereby, the following technical advantage is achieved: In many countries and / or regions, warning signs are attached to the rear of vehicles with protruding loads, and if such a warning sign is identified, the result of the object identification can be verified, thus improving the reasonableness and accuracy of the result of the identification.

[0028] Optionally, the step S2 further comprises:

[0029] During capturing of the second image of the object, the distance variation of the monocular camera to the object is kept at a stable value.

[0030] Thereby, the following technical advantage is achieved: It is essential for the quality of the stereo vision analysis that the longitudinal distance variation of the monocular camera to the target object is known as accurately as possible. Therefore, any longitudinal distance fluctuations during capturing / detecting can lead to inaccurate estimation results. By keeping the distance variation substantially stable during capturing / detecting, the introduction of uncertainties can be kept as low as possible.

[0031] According to a second aspect of the present application, a device for determining a distance of a vehicle to an object by means of a monocular camera arranged on the vehicle is provided, the device being configured to perform the method according to the first aspect of the present application, the device comprising:

[0032] an acquisition module configured to acquire a first image and a second image of an object located in front of the vehicle captured by means of the monocular camera;

[0033] a control module configured to vary the distance of the monocular camera to the object along a driving direction of the vehicle within a time interval between capturing the first image and the second image by means of the monocular camera; and

[0034] an analysis module configured to perform a stereo vision analysis of the object based on the first image and the second image and configured to determine the distance of the vehicle to the object depending on a result of the stereo vision analysis.

[0035] According to a third aspect of the present application, a vehicle is provided, the vehicle comprising a device according to the second aspect of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0036] The principles, features and advantages of the present application can be better understood by reference to the following detailed description, taken in conjunction with the accompanying drawings, in which:

[0037] Figure 1 a block diagram of a vehicle comprising a device for determining a distance of the vehicle to an object by means of a monocular camera according to an exemplary embodiment of the present application is shown;

[0038] Figure 2 a flow chart of a method for determining a distance of a vehicle to an object by means of a monocular camera according to an exemplary embodiment of the present application is shown;

[0039] Figure 3 a schematic diagram showing the use of the method according to the present application in an exemplary application scenario is shown;

[0040] Figure 4a schematic diagram showing the use of the method according to the application in another exemplary application scenario; and

[0041] Figure 5 a schematic diagram showing the determination of the distance of a vehicle to a protrusion on an object by means of a motion stereo method. DETAILED DESCRIPTION

[0042] In order that the technical problems addressed by the present application, the technical solutions and the advantageous technical effects are made more apparent, in the following the present application will be further described in detail in conjunction with the drawings and a plurality of exemplary embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but are not used to limit the protection scope of the present application.

[0043] Figure 1 a block diagram of a vehicle according to an exemplary embodiment of the present application, the vehicle comprising a device for determining the distance of the vehicle to an object by means of a monocular camera.

[0044] As Figure 1 is shown, the vehicle 1 comprises a device 10 according to the present application. The vehicle 1 can for example partially autonomously travel or also completely autonomously travel. In order to be able to realize an accurate detection of the distance of the vehicle to an object by means of a monocular camera, the device 10 comprises an acquisition module 20, a control module 30 and an analysis module 40.

[0045] The acquisition module 20 is used to acquire an image sequence of an object located in front of the vehicle, which is taken by means of a monocular camera, the image sequence comprising a first image and a second image taken in time succession to each other. In this exemplary embodiment, the acquisition module 20 is directly configured as a monocular camera and is thus directly usable for taking images of the road surface in front of the vehicle. It is however also possible that the acquisition module 20 is configured as a communication interface and is connected in a wired and / or wireless manner to a monocular camera arranged on the vehicle 1 in order to be able to receive the taken images from the monocular camera.

[0046] The control module 30 is configured to change the distance of the monocular camera to the object along the driving direction of the vehicle within the time interval between the taking of the first image and the second image by means of the monocular camera. To this end, the control module 30 is connected, for example, to the movement mechanism 110 of the monocular camera and to the accelerator pedal 21 and the brake pedal 22 of the vehicle 1, respectively. Thereby, the control module 30 is able to send control signals to the movement mechanism 110 of the monocular camera, for example, after the taking of the first image, in order to cause the movement mechanism 110 to move the monocular camera relative to the vehicle 1, so as to change the distance of the monocular camera to the object in front along the driving direction of the vehicle. Furthermore, the control module 30 is also able to send control signals to the accelerator pedal 21 or the brake pedal 22 of the vehicle 1, for example, after the taking of the first image, in order to cause the vehicle 1 to accelerate or decelerate along the original driving direction, so as to change the distance of the vehicle 1 (and thus of the monocular camera) to the object in front. It is also possible that the control module 30 is connected to other mechanisms of the vehicle 1 for controlling the vehicle speed or to a function module for distance control, which is not limited by the present application.

[0047] The analysis module 40 is connected to the acquisition module 20, for example, in order to receive the first image and the second image taken by means of the monocular camera therefrom. Furthermore, the analysis module 40 is connected to the control module 30, so as to be able to receive information about the change of the longitudinal distance of the monocular camera to the object from the control module 30 in real time and to use it for the image analysis process. A plurality of image recognition and processing algorithms are integrated in the analysis module 40, for example, by means of which a stereovision analysis of the object in front of the vehicle 1 can be carried out in the analysis module 40 on the basis of the first image and the second image. On the basis of the results of the stereovision analysis, a reliable determination of the distance of the vehicle 1 to the object in front can be achieved by means of appropriate calculations.

[0048] Figure 2 A flow chart of a method for determining the distance of a vehicle to an object by means of a monocular camera according to an exemplary embodiment of the present application is shown. The method can be carried out, for example, in the case of a device 1 according to Figure 1 as shown.

[0049] The method comprises a step S1 in which a first image of an object in front of the vehicle is taken by means of a monocular camera. The monocular camera is arranged, for example, on the roof of the vehicle or inside the driver's cabin of the vehicle. Such a monocular camera can be fixed relative to the vehicle, however, it is also possible that the monocular camera is mounted on the vehicle in such a way that it is movable relative to the vehicle.

[0050] After the first image has been taken, the object in front of the vehicle is identified in the first image, for example in optional step S11 by means of a trained classifier and / or artificial neural network. Here, such a classifier and / or artificial neural network can be trained beforehand, for example by means of a large number of photos with annotation information or artificially created images, and after the training has been completed, an image taken by means of a monocular camera can be input into the classifier and / or artificial neural network, so that, for example, features such as the class, size, contour, dimensions, etc. of different traffic objects in the taken image can be derived. Here, if a vehicle with a protruding load is required to be actively identified in a particular country and / or region, a plausibility check of the identified result can also be carried out by means of typical identifiers and / or labels on the object, such as flags, warning signs, barcodes, etc. The result of such an identification can be stored in a memory in order to be available for subsequent analysis processes.

[0051] In optional step S12, the height of the protrusion on the object in front of the vehicle relative to the ground can also be identified on the basis of the first image. Here, although the 3D contour information and the associated depth information of the object cannot be precisely derived from the single taking result by means of the monocular camera, what has already been achieved on the basis of such a 2D detection result is that the protrusion on the object in front is identified by means of an image recognition algorithm on the basis of color, texture, shape, etc. information, but at this point the degree of protrusion of the protrusion is still unknown, and then the height of any point on the protrusion relative to the ground is calculated by means of a suitable coordinate transformation algorithm.

[0052] In optional step S13, the height of the protrusion relative to the ground is compared with the height of the vehicle itself, and it is determined whether the protrusion is higher than the height of the vehicle itself. Here, some basic information about the own vehicle, such as the height of the own vehicle, the position of the center of gravity, etc., is already stored, for example as pre-known information, in the vehicle locally and / or in a cloud server and can be called up when required.

[0053] If it is determined in step S13 that the protrusion is higher than the height of the vehicle itself, the protrusion can be excluded from the distance determination process, for example in step S14. The protrusion being higher than the height of the vehicle itself means, for example, that the threat to the vehicle itself in the event of a collision is small, so that in this case the structural feature can not be taken into account when performing the distance determination. As an example, the distance detection process according to the application can be continued after step S14 (for example, step S2 is continued). As another example, a distance detection method based on another principle can also be performed after step S14, in which, for example, the steps of taking a second image and generating a longitudinal distance change can be omitted, so that the time and calculation effort is saved to the greatest extent.

[0054] If it is determined in step S13 that the protrusion is below the vehicle's own height, then, for example, the distance of the monocular camera to the object is first changed in the driving direction of the vehicle in step S2, and then a second image of the object in front of the vehicle is taken by means of the monocular camera. By way of example, the vehicle can be caused to change the distance to the object in the driving direction by suitable acceleration or deceleration. By way of further example, the monocular camera can also be caused to move relative to the vehicle in the longitudinal direction of the vehicle, for example by means of a movement mechanism, so that the distance of the monocular camera to the object can likewise be changed. During the taking of the second image, the amount of change in the distance of the monocular camera to the object can be kept constant, for example by controlling the vehicle speed or by controlling the position of the monocular camera relative to the vehicle, so as to ensure the reliability of the subsequent analysis process.

[0055] Next, in step S3, a stereovision analysis of the object in front of the vehicle is carried out on the basis of the first and second images, and the distance of the vehicle to the object is determined on the basis of the result of the stereovision analysis. Here, since a difference in the angle of view for the monocular camera in the driving direction of the vehicle is created for taking the image of the object in front, it is possible, for example, to reconstruct 3D profile information of the object in front of the vehicle by means of the method of motion recovery of structure, on the basis of which the distance of the vehicle to the protrusion on the object can be determined. The distance determined in the method according to the application is more precise than in the case of a distance measurement on the basis of a 2D image, so that it can be used, for example, to control the distance-keeping function and the obstacle avoidance function of the vehicle more reliably.

[0056] Figure 3 A schematic diagram showing the use of the method according to the application in one exemplary application scenario is shown.

[0057] In the scenario shown, a vehicle 1 comprising a monocular camera 11 is driving in a defined driving direction, in front of which there is a lorry 2 with a protruding structure 3. In the sense of the application, "in front of the vehicle" does not mean only that the object is located completely in the driving direction of the vehicle or in other words is positioned on the prolongation of the longitudinal axis of the vehicle, but also includes the case in which the lorry 2 is offset in the longitudinal direction relative to the vehicle 1, while there is also a transverse offset. The protruding structure 3 on the lorry 2 can be, for example, a towed load, a trailer or a relief decoration attached to the lorry 2, but can also be an integral part of the appearance of the lorry 2.

[0058] As Figure 3 As shown in the upper part, in order to determine the exact distance of the vehicle 1 to the lorry 2 in front, i.e. for example in order to determine the distance of the vehicle 1 to the protruding structure 3 on the lorry 2, a first image of the lorry 2 is first taken by means of the monocular camera 11 in a first position POS 1 of the vehicle 1 relative to the lorry 2. Here, this first position POS 1 is understood to be any position in the course of driving of the vehicle 1 in the current driving direction. Here, the object in front is observed and detected by means of the monocular camera 11 with a defined field of view E1.

[0059] Based on the first image, it can be recognized by means of a trained machine learning classifier and / or artificial neural network that the object in front is a truck 2 with a protruding structure 3. At the same time, it is also recognized that the height h of the protruding structure 3 relative to the ground is less than the height of the vehicle itself, which indicates a greater danger of the structure and which should therefore be taken into account in the distance detection process.

[0060] As Figure 3 The lower part shows that the vehicle 1 is then maneuvered to keep the original driving direction and to change the vehicle speed, so that a second position POS 2 of the vehicle 1 is formed. For ease of comparison, the first position POS 1 of the vehicle 1 is also shown in the lower part as a dotted line. Figure 3 It can be seen that, compared to the first position POS 1 of the vehicle 1, in the second position POS 2 the vehicle 1 is displaced by a displacement change Ad further away from the truck 2 in terms of driving behavior, which in turn leads to an increase in the distance of the monocular camera 11 on the vehicle 1 to the truck 2. In this case, a second image of the truck 2 is taken by means of the monocular camera 11 with a new field of view E2. Since there is a difference between the two fields of view E1, E2 with which the truck 2 is imaged by means of the monocular camera 11, a stereo vision analysis of the 3D profile of the truck 2 can be carried out on the basis of the first and second images, and the distance of the vehicle 1 to the protruding structure 3 on the truck 2 can thus be determined.

[0061] Figure 4 A schematic diagram showing the use of the method according to the application in another exemplary application scenario is shown. In the scenario shown, a vehicle 1 comprising a monocular camera 11 is driving in a defined driving direction, in front of which there is a truck 2 with a protruding structure 3. When distance detection is carried out using the method according to the application, Figure 4 The difference to Figure 3 is that the monocular camera 11 is arranged to be movable relative to the vehicle 1 in the driving direction of the vehicle, i.e. in the direction of the longitudinal axis of the vehicle.

[0062] As Figure 4 The upper part shows that, first of all, a first image of the truck 2 is taken with the monocular camera 11 in a first position POS 1 relative to the truck 2. In this first position POS 1, the monocular camera 11 is brought, for example, to the very front end of the vehicle 1 in the driving direction of the vehicle 1, i.e. in the direction of the longitudinal axis of the vehicle 1. In this case, the monocular camera 11 detects the object in front with a first field of view E1. Based on the first image, it can be recognized that the object in front is a truck 2 with a protruding structure 3, and that the protruding structure 3 has a height above the ground which is less than the height of the vehicle 1 itself.

[0063] As Figure 4As shown in the lower part, the vehicle 1 is basically kept at a constant distance to the object in front (this can be done for example by a vehicle following function based on 2D vision perception), while the monocular camera 11 is moved in the driving direction of the vehicle 1 into a second position POS 2 relative to the vehicle 1 and then takes a second image of the object in front by means of the monocular camera 11. For ease of reference, the first image is shown in the upper part and the second image is shown in the lower part. Figure 4 In the lower part, the monocular camera 11 in the first position POS 1 is also shown by means of a dotted line. It can be seen that in this second position POS 2, the monocular camera 11 is further away from the truck 2 in front in the driving direction of the vehicle 1 by a displacement change Ad compared to the first position POS 1. This results in the monocular camera 1 being able to observe and take a picture of the truck 2 with a new field of view E2. Since there is a difference in the field of view E1, E2 by means of which the truck 2 is taken by the monocular camera 11, a stereo vision analysis of the 3D profile of the truck 2 can be carried out on the basis of the first and second images and the distance of the vehicle 1 to the protruding structure 3 on the truck 2 can thus be determined.

[0064] Figure 5 A schematic diagram of the determination of the distance of a vehicle to a protruding part on an object by means of the method of moving restoration of structure is shown.

[0065] In Figure 5 The rear view of the truck 2 driving in front of the vehicle is shown in the upper part, it can be seen that a protruding structure 3 is also installed at the rear of the container 4 of the truck 2, for example for towing other vehicles or goods in certain situations.

[0066] As Figure 5As shown in the lower part, a first image of the truck 2 has been taken by means of the monocular camera in the first position POS 1 of the vehicle with the first field of view E1 and a second image of the truck 2 has been taken by means of the monocular camera in the second position POS 2 of the vehicle with the second field of view E2. In order to reconstruct a three-dimensional profile of the rear side of the truck 2 on the basis of the two-dimensional image sequence, first the feature points in the first and second image can be matched. This matching comprises not only a one-to-one correspondence of all end points on the protruding structure 3 in both images, but also a matching of the projection points of the protruding structure 3 (as seen from the monocular camera) onto the rear side of the container 4 to each other. By means of this matching and in combination with a 2D object detection algorithm, for example the following information can be known: when the vehicle is in the first position POS 1, a first longitudinal distance dl of the vehicle in the driving direction relative to the rear side of the container 4 of the truck 2, a side length bl of the projection of the rear side of the protruding structure 3 of the truck 2 onto the container 4; when the vehicle is in the second position POS 2, a second longitudinal distance d2 of the vehicle in the driving direction relative to the rear side of the container 4 of the truck 2, a side length b2 of the projection of the rear side of the protruding structure 3 of the truck 2 onto the container 4; a change Δd between the first longitudinal distance dl and the second longitudinal distance d2; a side length b of the rear side of the protruding structure 3. On the basis of the above information, for example the protruding length l of the protruding structure 3 relative to the rear side of the container 4 and the distance d of the vehicle to the protruding structure can be calculated by means of the following equations:

[0067]

[0068] l = dl - d

[0069] Although specific embodiments of the present application have been described in detail, they are merely for the purpose of illustration and are not to be construed as a limitation on the present application. Various substitutions, alterations and modifications can be conceived by those skilled in the art without departing from the spirit and scope of the present application.

Claims

1. A method for determining a distance of a vehicle (1) to an object by means of a monocular camera (11), the object having a protrusion (3), the monocular camera (11) being arranged on the vehicle (1), the method comprising the following steps: S1: taking a first image of the object (2) located in front of the vehicle (1) by means of the monocular camera (11); S2: changing the distance of the monocular camera (11) to the object (2) along the driving direction of the vehicle (1), wherein the position of the monocular camera (11) relative to the vehicle (1) is kept unchanged, the vehicle (1) is decelerated along the original driving direction so as to increase the distance of the vehicle (1) to the object in front, and then a second image of the object (2) is taken by means of the monocular camera (11), wherein the driving experience of a user is avoided from being affected in a significantly perceptible manner during the implementation of the scheme; and S3: performing a stereo vision analysis on the object (2) based on the first image and the second image, and determining the distance of the vehicle (1) to the protrusion (3) on the object (2) according to the result of the stereo vision analysis.

2. The method of claim 1, wherein, The step S3 comprises: reconstructing 3D contour information of the object (2) by means of a motion recovery structure method based on the first image and the second image, and determining the distance of the vehicle (1) to the protrusion (3) on the object (2) based on the 3D contour information.

3. The method of claim 1 or 2, wherein, The method further comprises the following steps before the step S2: identifying the height of the protrusion (3) on the object (2) relative to the ground based on the first image; comparing the height of the protrusion (3) relative to the ground with the height of the vehicle (1) itself; and ignoring the protrusion (3) when determining the distance when the height of the protrusion (3) relative to the ground is greater than the height of the vehicle (1) itself.

4. The method of any one of claims 1 to 3, wherein, The method further comprises the following steps before the step S2: identifying the object (2) in front of the vehicle (1) by means of a trained classifier and / or artificial neural network.

5. The method of any one of claims 1 to 4, wherein, The method further comprises the following steps: performing a credibility check on the result of the identification by means of a typical identifier and / or label on the object (2).

6. The method of any one of claims 1 to 5, wherein, The step S2 further comprises: keeping the distance changing amount of the monocular camera (11) to the object (2) at a stable value during taking the second image of the object (2).

7. An apparatus (10) for determining a distance of a vehicle (1) to an object (2) by means of a monocular camera (11), the monocular camera (11) being arranged on the vehicle (1), the apparatus (10) being configured to perform the method according to any one of claims 1 to 6, the apparatus (10) comprising: an acquisition module (20) configured to acquire a first image and a second image of the object (2) located in front of the vehicle (1) taken by means of the monocular camera (11); a control module (30) configured to change the distance of the monocular camera (11) to the object (2) along the driving direction of the vehicle (1) within a time interval between taking the first image and the second image by means of the monocular camera (11); and a determination module (40) configured to perform a stereo vision analysis on the object (2) based on the first image and the second image, and determine the distance of the vehicle (1) to the protrusion (3) on the object (2) according to the result of the stereo vision analysis. an analysis module (40) configured to enable a stereovision analysis of the object (2) based on the first image and the second image and to determine a distance of the vehicle (1) to the object (2) depending on a result of the stereovision analysis.

8. A vehicle (1) comprising an apparatus (10) according to claim 7.

Images