Method for calibrating an image sensor, device and associated motor vehicle.

Abstract

A calibration method for a first image sensor in a motor vehicle traveling in an environment, comprising the following steps: Receiving a first image (I1) of the environment from the first image sensor; Receiving a second image (I2) of the environment from a second image sensor, the second image sensor being fixed within the environment; Associating first geometric elements (I1_SEG300, I1_SEG400, I1_LIGN1) of the first image with second geometric elements (I2_SEG300, I2_SEG400, I2_LIGN1) of the second image; Determining first 3D coordinates of the first geometric elements (I1_SEG300, I1_SEG400, I1_LIGN1) from second 3D coordinates of the associated second geometric elements; Determining an orientation of the first image sensor of the motor vehicle.from the first 3-dimensional coordinates of the first geometric elements (I1_SEG300, I1_SEG400, I1_LIGN1). Figure for the abbreviation: figure 2,

Description

Title of the invention: Method for calibrating an image sensor, associated device and motor vehicle.

[0001] The invention relates to motor vehicles.

[0002] The increasing automation of motor vehicle driving requires precise perception of the vehicle's environment using image sensors. However, any error in the calibration knowledge, particularly in the knowledge of the image sensor orientation, can impact the accuracy of this perception, or even render it inoperative. This is the case, in particular, when the perception of the vehicle's environment is obtained from the fusion of images from several image sensors on the motor vehicle.

[0003] To remedy this drawback, the invention relates to a method for calibrating a first image sensor included in a motor vehicle, implemented while the motor vehicle is traveling in an environment surrounding the motor vehicle, the method being characterized in that it comprises the following first steps: • Reception, from the first image sensor, of a first image of a part of the environment, acquired by the first image sensor, then • Obtaining (in other words: detecting or determining) initial geometric elements from (in other words: within) the first image, • And, receiving, from a second image sensor, a second image of the part of the environment, acquired by the second image sensor, the second image sensor being contained within the environment and fixed relative to the Earth's surface (in other words: fixed relative to the ground), then • Obtaining (in other words: Detection or determination) second geometric elements from (in other words: within) the second image, obtaining the second geometric elements including determining the second 3-dimensional coordinates of each second element of the second geometric elements, • Following the step of obtaining the first geometric elements and the step of obtaining the second geometric elements, each first geometric element of the first set of geometric elements is associated with a (different) element of the second set of geometric elements. • Determination of initial 3-dimensional coordinates for each first element of the first geometric elements from the second 3-dimensional coordinates of the element (in other words: of an element) of the second geometric elements associated with said first element during the association step, • Then, determination of an orientation of the first image sensor of the motor vehicle, from the first 3-dimensional coordinates of each first element of the first geometric elements.

[0004] Thus, thanks to the invention, the first image sensor can be calibrated during the movement of the motor vehicle on a public road, using the second sensor, without requiring the motor vehicle to be driven to a place dedicated to such calibration.

[0005] For example, the second 3-dimensional coordinates of each second element of the second geometric elements are obtained from initial data including, for example: • A location and / or orientation of the second image sensor (known prior to the initial steps), and / or • Characteristics (e.g., geographical location, slope) of an element of the environment, for example of the roadway or of a marking on the ground (known prior to the first steps) for example when all or part of the second geometric elements touch the element of the environment.

[0006] Alternatively: • The second geometric elements are fixed and their coordinates are known and fixed (this may be the case for second elements obtained from fixed objects), • The second 3-dimensional coordinates are obtained directly by a radar or a laser remote detector.

[0007] The second image sensor is understood to be fixed relative to the Earth's surface (in other words: relative to the ground) by the fact that, for example, the second sensor is fixed (for example, welded, screwed, glued, or sealed) directly to the Earth's surface, or indirectly to the Earth's surface (in other words: to the ground), for example, via at least one of the following: • A building, • A road sign (fixed to the ground) • Street furniture, • Vegetation whose roots are in the ground (for example, a tree), • A post.

[0008] The vehicle is preferably in motion. Alternatively, it is stationary, by example at a stop sign or a traffic light.

[0009] It will be apparent to the person skilled in the art that at some point in the process, coordinates or data in a second fixed frame of reference relative to the ground (in other words: relative to the second image sensor) are converted into coordinates or data in a first fixed frame of reference relative to the motor vehicle (when it is in motion).

[0010] For example, all the first steps can be implemented in the second frame, and a conversion of the orientation in the first frame can then be carried out.

[0011] Conversely, the initial data can be converted in the first frame of reference. Thus, all the initial steps can be implemented in the first frame of reference.

[0012] This conversion can alternatively be carried out at intermediate stages of the process. For example, the step of determining the first three-dimensional coordinates can include such a conversion. Alternatively, the determination of the first three-dimensional coordinates can be implemented directly. That is, the first three-dimensional coordinates of each first element of the first geometric elements can be a copy of the second coordinates of the second element associated with said first element during the association step.

[0013] In a known manner: • This conversion includes a step of determining the relative position of the first coordinate system with respect to the second coordinate system, and • Such a relative position can be determined from a geographical location and an orientation of the motor vehicle.

[0014] The method can thus include a step of receiving a geographical location and an orientation of the motor vehicle (for example from a satellite geolocation module of the motor vehicle, for example of the so-called "GPS" type, or alternatively from a detector of the road infrastructure).

[0015] Alternatively, one or more wheels of the motor vehicle are positioned (temporarily) against a movable wedge so that the geographical location and orientation of the motor vehicle remain constant. According to this alternative, the second 3D coordinates can be directly expressed in the first coordinate system, without conversion.

[0016] The method, according to the invention, may include: • An environmental analysis step (i.e., detection and / or recognition and / or estimation of the distance of objects in the environment) based on images received from the first image sensor, taking into account the orientation of the first image sensor, A step of ordering a component of the motor vehicle (for example: a propulsion motor, a brake, a screen or a projector) from a result of this analysis step.

[0017] During the step of determining the orientation of the first image sensor of the motor vehicle, a location of the first image sensor, for example in the first reference frame, can be considered predetermined, the accuracy of this position having a minor influence on the analysis of the environment. Alternatively, the step of determining the orientation of the first image sensor of the motor vehicle includes determining a location of the first image sensor, for example in the first reference frame.

[0018] The orientation may include, for example, a roll angle, a pitch angle, a yaw angle of the image sensor.

[0019] A person skilled in the art knows how to implement the step of determining the orientation of the first image sensor of the motor vehicle, based on the initial 3D coordinates of each of the first geometric elements. This step corresponds to a well-known problem in computer vision. By way of example, the following methods can be used: By solving a system of equations by singular value decomposition (SVD) and non-linear optimization, as for example described in the article: “LiDAR-camera calibration using line correspondences. Sensors”, Bai, Z., Jiang, G., & Xu, A. (2020), Sensors, 20(21), 6319. By supervised learning and training of a convolutional network, as for example described in the article: “CalibRCNN: Calibrating camera and LiDAR by recurrent convolutional neural network and geometry constraints”, Shi, J., Zhu, Z., Zhang, J., Liu, R., Wang, Z., Chen, S., & Liu, H. IEEE / RSJ International conference on Intelligent Robots and Systems, 2020, pp. 10197-10202, IEEE.

[0020] According to one embodiment: The first image sensor and the second image sensor are cameras (video), and The first image and the second image are 2-dimensional images.

[0021] According to a first variant, in one embodiment, the first image sensor is a laser remote detector, for example also called "LiDAR", and for example the second image sensor is also a laser remote detector, for example also called "LiDAR".

[0022] In this case, the invention can be implemented, among other things, for example, by the method described in the following publication: • “Pairwise LIDAR calibration using multi-type 3D geometry features in natural scene”. M He, H Zhao, F Davoine, J Cui, H Zha. 2013 IEEE / RSJ International Conference on Intelligent Robots and and Systems (IROS), 2013.

[0023] According to one embodiment: • The step of obtaining the second geometric elements includes a step of determining second bounding boxes, (images) of first moving objects (i.e., each of the second bounding boxes encompassing one of the first moving objects (in other words: a second image of one of the first moving objects)) in (in other words: from) the second image, at least part of the second geometric elements being obtained from the second bounding boxes, and • The step of obtaining the first geometric elements includes a step of determining first bounding boxes (of images) of the first moving objects (i.e. each of the first bounding boxes encompassing one of the first moving objects (in other words: a first image of one of the first moving objects)) in (in other words: from) the first image, at least part of the first geometric elements being obtained from the first bounding boxes.

[0024] Alternatively, other approaches are conceivable, for example, using the pixels of the first and second images.

[0025] The step of determining the second bounding boxes can be implemented by an artificial intelligence model, for example by an electronic neural network. Those skilled in the art are familiar with various methods for implementing such a step. Examples include the methods described in the following articles: • By supervised learning of a convolution network, described for example in the article: “3D Bounding box detection from monocular images”, Catà Villà, M., Master's thesis, Universitat Politècnica de Catalunya. 2019, • Through learning that includes geometric constraints, described for example in the article: “Deep monocular 3d object detection with closed-form geometry constraints” Naiden, A., Paunescu, V., Kim, G., Jeon, B., & Leordeanu, 2019 IEEE international conference on image processing (ICIP) (pp. 61-65).

[0026] The step of determining the first encompassing boxes is quite classic in the automotive field, and can be implemented by an artificial intelligence model, for example by an artificial neural network.

[0027] According to one embodiment, the first moving objects are pedestrians. Alternatively, for example, they are motor vehicles or cyclists.

[0028]

[0029]

[0030]

[0031]

[0032]

[0033]

[0034]

[0035] According to one embodiment: • The second set of encompassing boxes are: • Rectangular parallelepipeds (or cuboids), or • Cylinders, (or alternatively, for example right prisms). • The first encompassing boxes are rectangles. For example, in this embodiment: • The second set of encompassing boxes each comprises two horizontal bases, • The first bounding boxes are rectangles comprising two horizontal segments. Alternatively, for example, the second encompassing boxes are prisms, cones or spheres, and the first encompassing boxes are trapezoids, triangles or circles respectively. In one embodiment, the second and first geometric elements comprise (or are) segments. In this case, for example, the first three-dimensional coordinates of each first element of the first geometric elements and / or the second three-dimensional coordinates of each second element of the second geometric elements can be the coordinates of the endpoints of the segments. For example : • Each of the two horizontal bases includes a center, • Each of the two horizontal segments of the first bounding boxes includes a center. According to one embodiment: • The second geometric elements include a second vertical segment (i.e., parallel to the direction of gravity) connecting the centers of the two horizontal bases (i.e., connecting the center of one of the two horizontal bases to the center of the other of the two horizontal bases) of one of the second encompassing boxes, • The first geometric elements include a first vertical segment connecting the centers of the two horizontal segments (i.e.: parallel to the Earth's horizon) (i.e.: connecting the center of one of the two horizontal segments to the center of the other segment of the two horizontal segments) of one of the first encompassing boxes. Alternatively, the first and second bounding boxes include edges, the segments are these edges (vertical and / or horizontal). Alternatively, the first geometric elements include vertices of the first bounding boxes and the second geometric elements include vertices of the second bounding boxes. The association step then becomes a well-known problem called "Perspective-N-Point" for which the person skilled in the art knows several approaches.

[0036] In the case of the first geometric elements, a person skilled in the art can easily implement treatments to distinguish vertical segments from horizontal segments.

[0037] For example, an approximate direction of the horizontal and vertical can be known (an orientation, admittedly approximate, which the method according to the invention specifies, of the first image sensor being known prior to the implementation of the first steps). A segment whose direction is closer to an approximate horizontal direction than to an approximate vertical direction is determined to be horizontal. A segment whose direction is closer to an approximate vertical direction than to the horizontal direction is determined to be vertical.

[0038] Alternatively, the first vertical segment is determined by the artificial intelligence model, for example by the artificial neural network which determines the first bounding boxes.

[0039] The horizontal and vertical segments can also be determined by an object recognition (a road, a pedestrian) or a dimension ratio between the vertical segments and the horizontal segments.

[0040] According to one embodiment: • The second geometric element detection step includes a detection step (of images) of a second line of ground markings in the second image, the second geometric elements comprising segments extending along the entire length of the second line of ground markings, and • The first geometric element detection step includes a detection step (of images) of a first line of ground marking in the first image, the first geometric elements comprising segments extending all along the first line of ground marking.

[0041] A person skilled in the art is familiar with various methods for implementing the step of detecting the first line of road markings. Examples include the methods described in the following article: • . “Recent progress in road and lane detection: a survey. », Hillel, AB, Lerner, R., Levi, D., & Raz, G. Machine vision and applications, 25(3), 727-745.

[0042] Regarding the step of detecting the second line of road markings, a person skilled in the art is familiar with various methods for implementing the detection step of Second lines of road markings. Examples include the following methods: • By supervised learning using a convolutional neural network, for example described in the following article: • “3D lanenet: end-to-end 3D multiple lane detection Garnett”, N., Cohen, R., Pe'er, T., Lahav, R., & Levi, D. Proceedings of the IEEE / CVF International Conference on Computer Vision pp. 2921-2930, 2019. • Through semi-supervised learning using Transformers, for example described in the following article: • “Performer: 3d lane detection via perspective transformer and the openlane benchmark”. Chen, L., Sima, C., Li, Y., Zheng, Z., Xu, J., Geng, X. & Yan, J.. European Conference on Computer Vision, 2022 October, pp. 550-567. Cham: Springer Nature Switzerland.

[0043] A person skilled in the art knows several solutions for implementing the association step, for example: • By a geometric search for the closest point between two lines to be matched, as described for example in the following article: • “An invariant, closed-form solution for matching sets of 3D lines”, Kamgar-Parsi, B., & Kamgar-Parsi, B. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Vol. 2, pp. 11-431. 2004 IEEE Computer Society, • By random sampling and probabilistic estimation (a family of Monte Carlo-type methods), for example using the so-called "RANSAC" algorithm, described in the following article: • “Camera pose estimation based on PnL with a known vertical direction”, Lecrosnier, L., Boutteau, R., Vasseur, P., Savatier, X., & Fraundorfer, F. IEEE Robotics and Automation Letters, 4(4), 3852-3859. 2019.

[0044] In fact, the association step associates each element of the first geometric elements obtained from a first detection of an object, in the first image, with an element of the second geometric elements obtained from a second detection of this object, in the second image.

[0045] According to one embodiment, the process includes second steps, the second steps being implemented by the motor vehicle, the second steps comprising: • Before the association step, there is a receiving step (and a sending step) of the second image (from (in other words: originating from) the second image sensor) and, second geometric elements, via radio frequency communication (for example, of the so-called "V2X" type), • And, before the step of determining the first 3-dimensional coordinates, a step of receiving (and a step of sending) the second 3-dimensional coordinates of each second element of the second geometric elements (up to the motor vehicle or in other words, to the motor vehicle). • The following steps: • The reception stage, from the first image sensor, of the first image, • The stage of detecting the first geometric elements, • The association stage, • The step of determining the first 3-dimensional coordinates of each first element of the first geometric elements, • The step of determining the orientation of the first image sensor.

[0046] Other distributions of the steps are of course possible.

[0047] For example, the step of detecting the second geometric elements can be implemented in the motor vehicle (the initial data must, in this case, be received by the motor vehicle).

[0048] Alternatively, all the first steps can be implemented by the road infrastructure which can then communicate the orientation to the motor vehicle.

[0049] The invention also relates to a computer program comprising instructions, executable by a microprocessor or a microcontroller or a computer, for the implementation of the second steps, when executed by the microprocessor or the microcontroller or the computer.

[0050] The second steps can be implemented by an electronic device (or a motor vehicle). The invention therefore also relates to an electronic device (or a motor vehicle) configured to implement the second steps, as well as a motor vehicle comprising the electronic device.

[0051] The characteristics and advantages of the computer program, the electronic device, and the vehicle are identical to those of the method according to the invention (without it being necessary to repeat them here).

[0052] When the electronic device, the motor vehicle, (or other element) is "configured to" (or "capable of") performing or implementing a step or operation, this implies, for example, that the element includes means to to carry out the step or operation. The means preferably include electronic means, for example a computer program, data in memory, specialized electronic circuits, wired or wireless connections, a microprocessor and / or a microcontroller.

[0053] Other features and advantages of the present invention will become more apparent upon reading the following detailed description, which includes embodiments of the invention given by way of non-limiting examples and illustrated by the accompanying drawings, in which: • [Fig. 1] represents an electronic device and a motor vehicle, according to an embodiment of the invention, in top view, • [Fig. 2] represent images obtained during the implementation of the process during steps of the process of [Fig.3]. • [Fig.3] represents an implementation of the process according to the invention.

[0054] In [Fig.1], certain elements are, of course, seen through transparency.

[0055] Detailed description of an example embodiment of the invention, with reference to figures 1 to 3.

[0056] Figure 1 represents a vehicle 100, which is a motor vehicle. The vehicle 100 comprises a microprocessor 110 connected (by connections represented by a solid line) to the following components of the vehicle 100: • A screen 140, • A 120 camera, and • A satellite geolocation module 130 (for example of the so-called "GPS" type).

[0057] With reference to [Fig.3], during stages SOObis to S40bis and S00 to S70, the vehicle 100 travels on a public road 600, in an environment envlOO which includes a tree 500, and a camera 200 including a microprocessor 210.

[0058] The camera 200 is fixed relative to the ground 700. It is for example bolted to a post (not shown) itself sealed to the ground 700.

[0059] At step S00, the microprocessor 110 receives, from the camera 120, a first image II, [Fig.2], of a part plOO of the environment envlOO, acquired by the camera 120.

[0060] The plOO part of the envlOO environment includes pedestrians 300, pedestrian 400 and a MARQ1 ground marking.

[0061] At the SOObis stage, the microprocessor 210 receives, from the camera 200, an image 12, represented [Fig.2], of the part plOO of the environment envlOO, acquired by the camera 200.

[0062] For example, steps S00 and SOObis are implemented simultaneously. Alternatively, steps S00 and SOObis are separated by a period of time taken into account during the association stage.

[0063] At step S10, the microprocessor 110 determines the bounding boxes I1_BOX300 and I1_BOX400 of pedestrians 300 and 400 respectively in image II.

[0064] At the SlObis stage, the microprocessor 210 determines the bounding boxes I2_BOX300 and I2_BOX400 of pedestrians 300 and 400 respectively in image 12.

[0065] For example, the bounding boxes I1_BOX300,11_BOX400,12_BOX300 and I2_BOX400 are determined according to the methods mentioned above.

[0066] At step S20, the microprocessor 110 determines the segments Il_SEG300 and Il_SEG400 linking the centers of two horizontal segments of the bounding boxes I1_BOX300,11_BOX400.

[0067] At step S20bis, the microprocessor 210 determines the segments I2_SEG300 and I2_SEG400 linking the centers of two horizontal bases of the bounding boxes I2_BOX300 and I2_BOX400.

[0068] At step S30, the microprocessor 110 detects the ground marking line I1_MARQ1, and determines the segment I1_LIGN1 extending along the ground marking line I1_MARQ1, according for example to one of the methods mentioned above.

[0069] At step S30bis, the microprocessor 210 detects the ground marking line I2_MARQ1, and determines the segment I2_LIGN1 extending along the entire length of the ground marking line I2_MARQ1.

[0070] At step S40bis, the microprocessor 210 determines the 3-dimensional coordinates of the segments I2_SEG300,12_SEG400,12_LIGN1, for example of the vertices of the segments I2_SEG300,12_SEG400,12_LIGN1, in the frame rep2.

[0071] These 3-dimensional coordinates are obtained from a location and orientation of the camera 200 in the reference frame rep2, for example in the memory of the microprocessor 210.

[0072] The characteristics of the roadway 600 and the ground 700 surrounding it, for example in the memory of the microprocessor 210 (for example, its altitude, its geographical coordinates, its slope, in the frame 2) are also used to determine not only the segment I2_LIGN1, but also the segments I2_SEG300 and I2_SEG400, because the pedestrians are located directly on the ground 700 (i.e.: the bottom of the segments I2_SEG300 and I2_SEG400 touches the ground).

[0073] At step S40, the segments I2_SEG300,12_SEG400,12_LIGN1, the 3-dimensional coordinates of the segments I2_SEG300,12_SEG400,12_LIGN1, as well as the image 12, are transmitted, by radio frequency communication, for example of the type called "V2X", by the microprocessor 210 to the microprocessor 110 which receives them.

[0074] At step S50, the microprocessor 110 combines: • Segment I1_SEG300 to segment I2_SEG300, • Segment II_SEG400 to segment I2_SEG400, • The segment I1_LIGN1 to the segment I2_LIGN1.

[0075] This association step can be implemented by one of the methods mentioned above, known to a person skilled in the art.

[0076] At step S50, the microprocessor 110 converts to a fixed reference frame relative to For vehicle 100, the 3D coordinates of the segments:

[0077] Il_SEG300 equal to the coordinates of segment I2_SEG300,

[0078] Il_SEG400 equal to the coordinates of segment I2_SEG400,

[0079] I1_LIGN1 equal to the coordinates of segment I2_LIGN1.

[0080] Step S50 includes, for this purpose, a step of determining a relative position of the reference frame rep1 with respect to the reference frame rep2, from a geographical location and an orientation of the vehicle 100 coming from the satellite geolocation module 130 (and, of course, from the reference frame rep2).

[0081] At step S60, the microprocessor 110 determines an orientation of the camera 120, in the repi frame, from the 3-dimensional coordinates of the segments Il_SEG300, Il_SEG400 and I1_LIGN1, using one of the methods mentioned below.

[0082] At step S70, the microprocessor 110 analyzes the environment from images received from camera 120 and controls screen 140 based on a result of this analysis.

Claims

Demands

1. A method for calibrating a first image sensor (120) included in a motor vehicle (100), implemented while the motor vehicle (100) is traveling in an environment (envi00) surrounding the motor vehicle (100), the method being characterized in that it comprises the following first steps: • Reception (S00), from the first image sensor (120), of a first image (II) of a part (plOO) of the environment (envlOO), acquired by the first image sensor (120), • Obtaining (S20, S30) the first geometric elements (Il_SEG300, Il_SEG400,11_LIGN1) from the first image (II), • Reception (SOObis), from a second image sensor (200), of a second image (12) of the part (plOO) of the environment (envlOO), acquired by the second image sensor (200), the second image sensor being included in the environment (envlOO) and fixed relative to the ground, • Obtaining (S20bis, S30bis) second geometric elements (I2_SEG300,12_SEG400,12_LIGN1) from the second image (12), obtaining the second geometric elements (I2_SEG300,12_SEG400,12_LIGN1) including a determination (S40bis) of second coordinates in 3 dimensions of each second element of the second geometric elements (I2_SEG300,12_SEG400,12_LIGN1), • Association (S50) of each first geometric element of the first geometric elements (Il_SEG300, Il_SEG400, I1_LIGN1) with an element of the second geometric elements (I2_SEG300,12_SEG400,12_LIGN1), • Determination (S60) of first 3-dimensional coordinates of each first element of the first geometric elements (Il_SEG300, Il_SEG400,11_LIGN1) from the second 3-dimensional coordinates of the element of the second geometric elements associated with said first element during the association step, • Then, determination (S70) of an orientation of the first image sensor (120) of the motor vehicle (100), from the first 3-dimensional coordinates of each first element of the first geometric elements (Il_SEG300, Il_SEG400, I1_LIGN1).

2. Calibration method according to the preceding claim wherein: • The first image sensor (120) and the second image sensor (200) are cameras, and • The first image (II) and the second image (12) are 2-dimensional images.

3. Calibration method according to the preceding claim, wherein: • The step of obtaining the second geometric elements (I2_SEG300, I2_SEG400, I2_LIGN1) comprises a step of determining (S10) second bounding boxes (I2_BOX400, I2_BOX300) of first moving objects (300, 400) in the second image (12), at least a portion of the second geometric elements (I2_SEG300, I2_SEG400, I2_LIGN1) being obtained from the second bounding boxes, and • The step of obtaining the first geometric elements (I1_SEG300, I1_SEG400, I1_LIGN1) comprises a step of determining (S10) first bounding boxes (I1_BOX300, I1_BOX400) of the first moving objects (I2_SEG300,12_SEG400,12_LIGN1) in the first image (II), at least part of the first geometric elements (II_SEG300, II_SEG400,11_LIGN1) being obtained from the first bounding boxes (I1_BOX300,11_BOX400), and

4. Calibration method according to the preceding claim, wherein: • The second encompassing boxes (I2_BOX400, I2_BOX300) are: • Rectangular parallelepipeds, or • Cylinders, and And, the first encompassing boxes (I1_BOX300, I1_BOX400) are rectangles.

5. Calibration method according to any one of the preceding claims wherein the second geometric elements and the first geometric elements comprise segments.

6. Calibration method according to any one of claims 3 to 5 wherein the first moving objects are pedestrians.

7. A calibration method according to any one of the preceding claims comprising second steps, the second steps being implemented by the motor vehicle (100), the second steps comprising: • Before the association step, a step of receiving the second image (12), and the second geometric elements (12_SEG300, 12_SEG400, 12_LIGN1), via radio frequency communication, • And, before the step of determining the first 3D coordinates of each first element of the first geometric elements (11_SEG300, 11_SEG400, 11_LIGN1), a step of receiving the second 3D coordinates of each second element of the second geometric elements (12_SEG300, 12_SEG400, 12_LIGN1), • The following steps: • The step of receiving, from the first image sensor (120), the first image (11),• The step of detecting the first geometric elements (Il_SEG300, Il_SEG400, Il_LIGN1), • The association step, • The step of determining the first 3D coordinates of each first element of the first geometric elements (Il_SEG300, Il_SEG400, Il_LIGN1), • The step of determining the orientation of the first image sensor (120).

8. A computer program comprising instructions, executable by a microprocessor or microcontroller, for the implementation of

9.

10. second steps according to claim 7, when executed by the microprocessor or microcontroller. Electronic device (110) configured to implement second steps according to claim 7. Motor vehicle (100) comprising the electronic device (110) according to the preceding claim.

Images