method of aligning a camera on a road

The method aligns a camera on a road using image processing to determine alignment parameters, addressing the challenge of tachometer reliance in existing systems and enabling cost-effective mobile radar device applications.

FR3131777B1Active Publication Date: 2024-11-29IDEMIA ROAD SAFETY FRANCE
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
FR2022000261
Authority / Receiving Office
FR · FR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2022-01-13
Publication Date
2024-11-29
Estimated Expiration
2042-01-13

AI Technical Summary

Technical Problem

Existing methods for aligning a camera in a road radar device require the use of a tachometer, which is cumbersome and expensive, making them unsuitable for mobile radar devices.

Method used

A method that determines camera alignment parameters using image processing to reduce the need for topographical surveys, involving the selection of pixels on road edges, calculation of parametric curves, and minimization of a score based on distance measurements to align the camera relative to a road model.

Benefits of technology

Enables precise camera alignment on a road without the need for a tachometer, facilitating the use of mobile radar devices and reducing costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000018_0000
    Figure 00000018_0000
  • Figure 00000019_0000
    Figure 00000019_0000
  • Figure 00000019_0001
    Figure 00000019_0001
Patent Text Reader

Abstract

A method for determining the alignment parameters of a camera relative to a road, based on an image previously acquired by said camera and a straight road model, the image showing a first edge of the road and a second edge opposite the first edge. The method comprises the steps of: selecting (104) pixels in the image located on a first edge of the road and several other pixels on a second edge of the road; from the selected pixels, determining (106) a first parametric curve along the first edge and a second parametric curve along the second edge; by projecting different points of the road model onto the image, determining, using the processing module, the camera alignment parameters by minimizing a score. Figure for the abstract: Fig. 2
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: method for aligning a camera on a road FIELD OF THE INVENTION

[0001] The present invention relates to a method for processing an image showing a road. This method finds advantageous application in the alignment of a camera of a road radar device, for example mobile. STATE OF THE ART

[0002] Today, the installation of a road radar device at a fixed location requires, if one wishes to obtain an accurate measurement of the speed of a vehicle traveling on a road, to precisely determine the position and orientation of the camera of the road radar device relative to this road.

[0003] In particular, document FR3096786 describes a method allowing such determination of position and relative orientation.

[0004] However, the method described in document FR3096786 has the disadvantage of requiring the use of a tachometer, which is not a very simple device to use and is also expensive. This method is therefore poorly suited to mobile radar devices. Statement of the invention

[0005] An aim of the invention is to align a camera with respect to a road by exploiting the information present in an image acquired by said camera and by minimizing the need to carry out topographical surveys.

[0006] To this end, according to a first aspect, a method is proposed for determining by a processing module alignment parameters of a camera relative to a road from an image previously acquired by said camera and a straight road model, the image showing a first edge of a road and a second edge of the road opposite the first edge, the straight road model being defined in a road reference frame by a first point, a second point, a third point and a fourth point, the first point and the third point defining a first limit of a first side of the road, the second point and the fourth point defining a second limit of a second side of the road opposite and parallel to the first side of the road, the first point and the second point defining a reference line perpendicular to the road, and the third point and the fourth point defining a line parallel to the reference line,the method comprising steps of: ,

[0007] - selection of pixels in the image, the selected pixels comprising several pixels on the first edge and several other pixels on the second edge,

[0008] - from the selected pixels, determination by the processing module in the plane of the image of a first parametric curve along the first edge and characterized by a first variable, and determination of a second parametric curve along the second edge and characterized by a second variable, each variable making it possible to calculate at any point of the corresponding parametric curve a tangent to said parametric curve,

[0009] - by projecting the first, second, third and fourth points of the model of route in the image, determination by the processing module of the camera alignment parameters by minimizing a score depending on the following distances:

[0010] • a distance between the projected in the plane of the image of the first point having co predefined ordinates in the road reference, and a point of the first parametric curve defined by a particular value of the first variable, called the first value,

[0011] • a distance between the projected in the plane of the image of the second point having predefined coordinates in the road reference frame, and a point of the second parametric curve defined by a particular value of the second variable, called the second value,

[0012] • a distance between a vanishing point observed in the image forming an intersection between a first tangent to the first parametric curve at the first point and a second tangent to the second parametric curve at the second point, and a reference vanishing point forming an intersection between the projected into the plane of the image of the first boundary of the road model and the projected into the plane of the image of the second boundary of the road model.

[0013] The method according to the first aspect may also comprise the following optional features, taken alone or in combination whenever technically possible.

[0014] Preferably, the score also depends on at least one of the following distances:

[0015] • a distance between a straight line of the image plane passing through the point of the first parametric curve defined by the first value and passing through the point of the second parametric curve defined by the second value, and a reference vanishing point forming an intersection between the projected into the image plane of the reference line of the road model, and the projected into the image plane of the line parallel to the reference line of the road reference.

[0016] • a distance between the first tangent and the projection in the plane of the image of the third point of the road model,

[0017] • a distance between the first tangent and the projection in the plane of the image of a fifth point of the first limit of the road model,

[0018] • a distance between the second tangent and the projection in the plane of the image of the fourth point of the road model,

[0019] • a distance between the second tangent and the projection in the plane of the image of a sixth point of the second limit of the road model.

[0020] Preferably, the third point and the fourth point define a first line parallel to the reference line, the straight road model being further defined by a fifth point located on the first boundary and a sixth point located on the second boundary, and the fifth point and the sixth point define a second line parallel to the reference line, the first and second parallel lines being located on either side of the reference line.

[0021] Preferably, the score also depends on at least one distance, a distance between a straight line of the image plane passing through the point of the first parametric curve defined by the first value and passing through the point of the second parametric curve defined by the second value, and a reference vanishing point forming an intersection between the projected in the image plane of the first parallel line and the second parallel line to the reference line of the road marker.

[0022] Preferably, the score depends on a sum of terms, each term being proportional to the cube of one of the distances.

[0023] Preferably, at least one of the first parametric curve and the second parametric curve is a B-spline, very preferably a B-spline of order 3.

[0024] Preferably, the method according to the first aspect comprises a step of polynomial smoothing of at least one of the first parametric curve and the second parametric curve, the smoothing being implemented before the calculation of the road model.

[0025] Preferably, the camera alignment parameters comprise at least one of the following parameters: an elevation angle of the camera relative to the road, a pitch angle of the camera relative to the road, an azimuth determined from the first value, the second value and the vanishing point intersection of the tangents, an image acquisition altitude relative to the road, a width of the road, a distance on the ground between the road and the camera having acquired the image.

[0026] Preferably, the minimization of the score is carried out on a discrete domain of the first value, the second value, the elevation angle, the pitch angle, the distance on the ground between the road and the camera, the acquisition altitude, the width of the road.

[0027] Also provided in a second aspect is a computer-readable memory storing computer-executable instructions for performing the steps of the method in accordance with the first aspect.

[0028] Also provided in a third aspect is a road control system comprising: a camera configured to acquire an image showing a first edge of a road and a second edge of the road opposite the first edge, a sensor, such as a Doppler radar or a lidar, configured to acquire data relating to the trajectory followed by a vehicle traveling on the road, the sensor being fixed relative to the camera, and a processing module configured to determine alignment parameters of the camera relative to the road by means of the method according to the first aspect. DESCRIPTION OF FIGURES

[0029] Other characteristics, aims and advantages of the invention will emerge from the following description, which is purely illustrative and non-limiting, and which must be read in conjunction with the appended drawings in which:

[0030] [Fig.l] schematically illustrates a road control system according to one embodiment.

[0031] [Fig. 2] is a flowchart of steps of a method according to one embodiment.

[0032] Figures 3, 4, 5b, and 6 are examples of images acquired during the method of [Fig. 2], in which certain points, straight lines and curves are highlighted.

[0033] [Fig.5a] illustrates a road marker.

[0034] Throughout the figures, similar elements bear identical references. DETAILED DESCRIPTION OF THE INVENTION 1) Road control system

[0035] With reference to [Fig.l], a road control system 1 comprises a camera 2, one or more motion sensors 4, and a processing module 6.

[0036] The camera 2 is capable of acquiring an image showing a first edge of a road and a second edge of the road opposite the first edge.

[0037] In this document, the term "camera" is to be interpreted broadly and covers any image acquisition device; in particular, the term "camera" covers but is not limited to a video acquisition device (in other words, image sequences).

[0038] The system 1 comprises a support 3 intended to be placed or fixed to the ground, the camera 2 being mounted on the support 3. The camera 2 can be fixed relative to the support 3, or else movable relative to the support 3, for example movable in rotation around an axis, preferably several axes (in which case the camera 2 is connected to the support by a ball joint). The support 3 is for example a tripod.

[0039] Generally, the motion sensor 4 is configured to acquire data relating to the trajectory followed by a moving target, such as a vehicle traveling on a road.

[0040] The motion sensor 4 may be a Doppler radar. The data relating to the trajectory of a target in this case include a radial velocity, i.e. a speed measured along a boresight axis separating the Doppler radar from the target. A speed of a target, such as a vehicle, can be deduced from an azimuth angle existing between the radar axis and the target's velocity vector corresponding to the position of that target.

[0041] Alternatively, the motion sensor 4 may be a lidar.

[0042] The system 1 further comprises a display screen 8, capable of displaying images acquired by the camera 2.

[0043] The system 1 further comprises an input interface 10 allowing a user to select pixels displayed by the display screen 8. The input interface 10 preferably comprises a touch screen coupled to the display screen 8, allowing a user to make such a selection by touching areas of the touch screen coinciding with areas of the display screen 8. Alternatively or additionally, the input interface 10 comprises a mouse and / or a keyboard.

[0044] The input interface 10 may also be configured to allow input of parameter values ​​by a user.

[0045] In one embodiment, the display screen 8 and the input interface 10 are part of a portable user terminal 12 independent of the camera 2. The terminal 12 takes the form, for example, of a tablet, a laptop computer or a smartphone.

[0046] The processing module 6 is, in general, configured to implement a processing method on the basis of an image acquired by the camera 2, processing which will be described in the remainder of this document. This processing method aims to determine alignment parameters of the camera relative to a road.

[0047] The processing module 6 comprises at least one memory storing the code instructions of a computer program, and at least one processor for executing this computer program, thus causing the implementation of this processing method.

[0048] The position of the processing module 6 in the system 1 is arbitrary. The processing module 6 can thus be a component:

[0049] • included in terminal 12 (as shown in [Fig.l]), or • attached to camera 2 or support 3, • included in a remote server communicating with the camera 2 and the display screen 8 via any communication network.

[0050] Alternatively, the processing module 6 comprises several processing units remotely from each other, communicating with each other, and implementing different steps of the processing method. These processing units can be distributed between the three aforementioned locations. It is thus understood that the steps of the method which will be described below are not necessarily implemented in the same location. 2) Road model

[0051] Data representative of a model of a road are stored in the memory of the processing module 6. This model assumes that the road is straight, that is to say that the modeled road has two opposite edges that are straight and parallel.

[0052] Let a reference frame be linked to the road (O, x, y, z). The reference frame is direct orthonormal and such that:

[0053] • the axis [Ox) is tangent to one of the two edges of the road, • the axis [Oy) is included in the road plane, • the axis [Oz) is orthogonal to the plane of the road, • point O is such that the camera has a ground position with coordinates (-L1, d), where L1 is a predefined constant, for example 28 meters.

[0054] The road model is defined by a plurality of points U1, U2, VI, V2, W1 and W2 having coordinates expressed in the reference frame linked to the road. These coordinates are as follows:

[0055] • U1: (-l_Size, 0, 0) • VI: (0, 0, 0) • W1: (1_Size, 0, 0) • U2: (-l_Size, -L, 0) • V2: (0, -L, 0) • W2: (1_Size, -L, 0)

[0056] where 1_Size is a predefined constant, for example equal to 10 meters and L corresponds to the width of the road.

[0057] Thus: point VI and point U1 define a first limit of a first side of the road and corresponding to a line of zero abscissa of the road reference frame; point V2 and point U2 define a second limit of a second side of the road opposite and parallel to the first side of the road; points VI and V2 define a reference line corresponding to a line of zero ordinate of the road reference frame perpendicular to the road; and points U1 and U2 defining a line parallel to the reference line.

[0058] Point V1 is located at the origin of the reference frame.

[0059] This choice of route reference is an example. The origin of the reference, the positioning of the abscissa and ordinate axes and / or their direction can be chosen differently.

[0060] The road model is further defined by a rotation and a translation. 3) Alignment process

[0061] The system 1 can typically be used for monitoring compliance with the highway code, for example by representatives of law enforcement agencies (police officers, gendarmes, etc.).

[0062] In the following, a method implemented by the system 1 in this application is detailed. particular application, while noting that other applications of the system can be considered.

[0063] A user of the system 1 is stationed near the edge of a road. The road comprises a first edge and a second edge opposite the first edge. The road comprises one or more traffic lanes between the first edge and the second edge. Each of the first edge and the second edge runs along a traffic lane.

[0064] The user positions the support on the ground next to the road, for example a few meters from the road. The camera 2 is oriented to be able to observe the road (more precisely a portion of this road). In particular, it is ensured that the first edge and the second edge of the road are all in the field of vision of the camera 2.

[0065] The method comprises a preliminary phase during which the position and orientation of the camera 2 relative to the road are determined.

[0066] With reference to [Fig.2], this preliminary phase comprises the following steps.

[0067] In an acquisition step 100, an image is acquired by the camera 2. Due to its orientation and its position relative to the road, the acquired image shows the first edge and the second edge of the road.

[0068] In a step 102, the processing module 6 controls the display of the image on the display screen 8.

[0069] In a selection step 104, pixels of the image located on the first edge and on the second edge as represented in this image are selected.

[0070] In a manual embodiment of step 104, such a selection is made by the user by means of the input interface 10. For example, the user can touch different areas of the touch screen; a pixel of the displayed image corresponding to each touched area is identified by the processing module 6, this pixel being for example aligned with the center of the touched area.

[0071] In an automatic embodiment of step 104, the selection of pixels is carried out automatically by the processing module 6, by means of a shape or contour recognition tool adapted to detect the first edge and the second edge in the image.

[0072] The processing module 6 can control the display of markers making it possible to highlight the pixels selected in the manual production mode or the automatic production mode.

[0073] In a semi-automatic embodiment of step 104, the two preceding principles are combined: the processing module 6 can control a display of markers highlighting pixels having been selected using the shape or contour recognition tool, so that these pixels can be located by the user, and can offer the user the possibility of adjusting the positions of these pixels. pixels, for example by dragging and dropping. This is advantageous because the image edge recognition tool may not work perfectly under certain conditions (for example, in the presence of shadows caused by vegetation near the road).

[0074] Preferably, N pixels located on the first edge and N pixels located on the first edge are selected in step 102, with N>1, for example N=6.

[0075] The coordinates of the pixels selected in the image plane are transmitted to the processing module 6 and are stored in the memory.

[0076] In a step 106, the processing module 6 determines, from the selected pixels, a first parametric curve and a second parametric curve in the plane of the image which run respectively along the first edge and the second edge of the road.

[0077] The first parametric curve curve_l may be a first B-spline. Such a B-spline is defined by parameters called in the literature “control points”, and by a variable tl with a value between 0 and 1, each value of the variable being associated with a point of the first B-spline.

[0078] Similarly, the second parametric curve curve_2 may be a second B-spline. Such a B-spline is defined by other control points, and by another variable t2 with a value between 0 and 1, each value of this other variable being associated with a point of the second B-spline.

[0079] The use of one or two B-splines is advantageous during the process, because it is easy to calculate a tangent to such a curve at any of its points (which is done later in the process).

[0080] In a particular embodiment, the or each B-spline used is of order 3, and N=6.

[0081] A B-spline calculated in step 106 may be subject to a wavelet phenomenon (or “snake” in English). To correct this phenomenon, an optional smoothing step 108 may be implemented, during which the processing module 6 applies polynomial smoothing to the or each determined B-spline.

[0082] The parameters defining the two parametric curves (if applicable smoothed at the end of the smoothing step 108) are stored in the memory.

[0083] By way of illustration, [Fig.3] shows an example of an image acquired during step 100, and on which two examples of parametric curves curve_l, curve_2 constructed from selected pixels have been plotted. The parametric curves curve_l, curve_2 are defined by variables tl, t2.

[0084] By way of illustration, [Fig.4] shows an example of an image acquired during step 100, and on which the tangents, named tangent_Tl and tangent_T2, considered at points Tl and T2. These points Tl and T2 are points belonging respectively to the parametric curves curve_l and curve_2 and defined by particular values ​​of the variables tl and t2, called respectively first value tl 1 and second value t21.

[0085] One of the properties of the reference frame linked to the road is that the axis [Ox) must be confused with the tangent line_Tl and that the axis [Oy) must be confused with the line (Tl T2).

[0086] We define Vx_B_SPLINES as the vanishing point of the tangents tangent_Tl and tangent_T2. We assume that the camera calibration matrix is ​​known.

[0087] In a step 110, the processing module 6 calculates a model of the road adapted to represent a straight road in the image from a definition of this road in the reference frame linked to the road.

[0088] In an embodiment corresponding to [Fig.5a],

[0089] • the camera has coordinates {-Ll, d, h] in this reference frame linked to the road

[0090] where Ll is a predefined constant, for example equal to 28 meters. The sign of d depends on the location of camera 2: shoulder, central reservation and the direction of view.

[0091] • the Euler angles characterizing the rotation are as follows: {azimuth + Pi / 2, elevation + Pi / 2, roll}.

[0092] One challenge is to find the alignment parameters of the camera 2 from an image acquired by said camera 2.

[0093] The alignment parameters of the camera 2 comprise at least one of the following parameters:

[0094] a. an elevation angle of the camera 2 relative to the road, b. a pitch angle of camera 2 relative to the road, c. an azimuth determined from the first value tl 1, the second value t21 and the vanishing point intersection of the tangents tangent_Tl, tangent_T2, d. an image acquisition altitude relative to the road, e. a width of the road L, f. a distance d on the ground between the road and camera 2 which acquired the image.

[0095] Determining the first and second values ​​tl 1, t21 for the variables tl and t2 allows the azimuth to be calculated. This is why in the remainder of the description, we seek to determine the parameters of first and second values ​​tl 1, t21 so that the azimuth alignment parameter can be determined.

[0096] We define:

[0097] • ul the projection of U1 in the image • u2 the projection of U2 in the image • vl the projection of VI in the image • v2 the projection of V2 in the image • wl the projection of W1 in the image • w2 the projection of W2 in the image • Vx the vanishing point defined by the intersection of the lines (ulwl) and (u2w2) • Vy the vanishing point defined by the intersection of the lines (ulu2) and (wlw2)

[0098] By way of illustration, [Fig.5a] represents the road model and [Fig.5b] represents the projection of this road model in an image acquired by the camera 2 for a given set of parameters {d, h, tll, t21, elevation, roll, L}. In [Fig.5b], the arrow pointing to the right means that the vanishing point Vy is located outside the image.

[0099] In a first variant, the aforementioned parameters are unknown. Concretely, the processing unit 6 makes the assumption that each unknown parameter of the road model is included in a predefined interval. The combination of these intervals thus covers a set of candidate road models.

[0100] For each candidate road model (defined by a septuplet of candidate values ​​for the parameters constituting unknowns), the processing module 6 calculates a score S associated with this candidate road model. The calculation of all possible scores can in practice be implemented in the form of a computer program comprising 7 loops nested within each other.

[0101] The road module determined in step 110 is the candidate road model which is associated with the minimum value score among all the calculated scores.

[0102] We will now detail how the score S associated with a candidate road model is calculated in step 110. The score depends on certain distances di measured in the image plane.

[0103] These distances di are defined below. The unit of distances is the pixel.

[0104] • dl: distance between ul and Tl • d2: distance between v2 and T2 • d3: distance between the vanishing points Vx_P_SPLINES and Vx • d4: distance between the vanishing point Vy and the line (Tl T2) • d5: distance between point ul and the tangent line_Tl • d6: distance between point wl and the tangent line_Tl • d7: distance between point u2 and the tangent line_T2 • d8: distance between point w2 and the tangent line_T2

[0105] Among the distances d1 to d8 defined above, the distances d4 to d8 are optional. In other words, the score S could depend only on the distances d1, d2 and d3. Nevertheless, taking them into account in the score S has the advantage of leading to the determination of a road model which is more likely to be reliable.

[0106] Generally speaking, the score S can depend on a sum of terms, each term of the sum being a function of one of the distances di defined above.

[0107] In a particularly advantageous mode, each term is proportional to the cube of one of the distances di. We can thus pose: 101081 s=E' =1 w'

[0109] or: 101101

[0111] where k is an integer equal to 3, 4, 5, 6, 7 or 8.

[0112] The inventor found that cubing the distances di for calculating the score S improves the alignment result compared to scores which would depend on the sum of the distances di, or even on the sum of the squares of these distances di.

[0113] As indicated previously, the road model determined in step 110 is the candidate road model associated with the score S of minimum value, among all the calculated scores.

[0114] In the first variant discussed so far, the parameters A, k, d, l1, / 2, elevation, and roll are unknown parameters of which we only know that their respective values ​​are included in predefined intervals.

[0115] The table below illustrates a predetermined and discretized search domain for the unknown parameters mentioned above (the search intervals and the search step). This domain covers 300,000 parameter combinations, or as many candidate road models. tll t21 elevation roll drive h L Interval center 0.4 0.6 0 0 3 1.6 3.5 Half interval 0.1 0.1 5 5 5 0 0.2 Number of values ​​per half interval 5 5 20 10 10 1 3 Steps 0.02 0.02 0.25 0.5 0.5 0 0.07

[0116] In a second variant, some of these parameters are predetermined, that is to say that their value is known before the implementation of step 110, and the other parameters are unknown. An advantage of this second variant is its shorter calculation time, because the number of candidate models is reduced (which translates in practice into a reduced number of nested calculation loops). In return, it is necessary to have been able to determine certain parameters in advance.

[0117] The or each predetermined parameter may be provided by the user by means of the input interface.

[0118] In particular, the parameters h, L and d are lengths which can be measured by the user with a measuring tool such as a tape measure.

[0119] The measurement of altitude h is easy to carry out and does not pose any safety problems even if it is variable since this measurement can be carried out at a distance from the road. On the other hand, the measurement of lengths L and d is more delicate since the user must approach and even cross the road to measure them.

[0120] This is why, in a preferred variant offering a good compromise between execution time and security, only the parameter h is measured by the user before the determination step 110, and the processing module 6 treats the other aforementioned parameters, including L and d, as unknowns during step 110.

[0121] After step 110, the processing module 6 can carry out in a step 112 a comparison between the minimum score S associated with the determined road model and a predefined threshold. If this minimum value of the score is lower than the predefined threshold, the determined road model is considered acceptable. If, on the other hand, this minimum value of the score is higher than the predefined threshold, the determined road model is not considered sufficiently reliable. In this case, the method returns to the selection step 102.

[0122] The threshold is for example 50 pixels.

[0123] Once a reliable road model has been determined by the processing module 6, the alignment process is completed. [Fig.6] illustrates a possible result of this process. The camera can be used for the application of road controls. This result was obtained after a calculation time of 90 seconds, with a score equal to 25 pixels.

Claims

1. Claims Method for determining by a processing module alignment parameters of a camera (2) relative to a road from an image previously acquired by said camera (2) and a straight road model, the image showing a first edge of a road and a second edge of the road opposite the first edge, the straight road model being defined in a road reference frame by a first point (V1), a second point (V2), a third point (U1) and a fourth point (U2), the first point (VI) and the third point (U1) defining a first limit of a first side of the road, the second point (V2) and the fourth point (U2) defining a second limit of a second side of the road opposite and parallel to the first side of the road, the first point (VI) and the second point (V2) defining a reference line perpendicular to the road, and the third point (U1) and the fourth point (U2) defining a line parallel to the reference line,the method comprising steps of:, - selection (104) of pixels in the image, the selected pixels comprising several pixels located on the first edge and several other pixels on the second edge, - from the selected pixels, determination (106) by the processing module in the image plane of a first parametric curve along the first edge and characterized by a first variable (tl), and determination of a second parametric curve along the second edge and characterized by a second variable (t2), each variable (tl, t2) making it possible to calculate at any point of the corresponding parametric curve a tangent to said parametric curve, - by projecting the first, second, third and fourth points (VI, V2, Ul, U2) of the road model into the image, determination by the processing module of the camera alignment parameters by minimizing a score depending on the following distances: - a distance between • the projection in the plane of the image of the first point (VI) having predefined coordinates in the road reference, and • a point (Tl) of the first parametric curve defined by a particular value of the first variable (tl), called the first value, a distance between • the projection in the plane of the image of the second point (V2) having predefined coordinates in the road reference, and • a point (T2) of the second parametric curve defined by a particular value of the second variable (t2), called the second value, a distance between • a vanishing point observed in the image (Vx_B_SPLINE) forming an intersection between a first tangent (tangent_Tl) to the first parametric curve at the first point (Tl) and a second tangent (tangent_T2) to the second parametric curve at the second point (T2), and • a reference vanishing point (Vx) forming an intersection between • the projected image plane of the first limit of the road model, and • the projected in the plane of the image of the second limit of the road model.

2. The method of claim 1, wherein the score also depends on at least one of the following distances: a distance between a straight line of the image plane passing through the point (Tl) of the first parametric curve defined by the first value and passing through the point (T2) of the second parametric curve defined by the second value, a reference vanishing point (Vy) forming an intersection between • the projected image plane of the reference line of the road model, and • the projected in the plane of the image of the line parallel to the reference line of the road marker. • a distance between the first tangent (tangent_Tl) and the projection in the image plane of the third point (Ul) of the road model, • a distance between the first tangent (tangent_Tl) and the projection in the image plane of a fifth point (Wl) of the first limit of the road model, • a distance between the second tangent (tangent_T2) and the projection in the image plane of the fourth point (U2) of the road model, • a distance between the second tangent (tangent_T2) and the projection in the image plane of a sixth point (W2) of the second limit of the road model.

3. Method according to one of the preceding claims, in which the third point (U1) and the fourth point (U2) define a first line parallel to the reference line, the straight road model being further defined by a fifth point (W1) located on the first boundary and a sixth point (W2) located on the second boundary, the fifth point (W1) and the sixth point (W2) define a second line parallel to the reference line, the first and second parallel lines being located on either side of the reference line, the score also depending on at least one distance a distance between • a straight line of the image plane passing through the point (T1) of the first parametric curve defined by the first value and passing through the point (T2) of the second parametric curve defined by the second value,• a reference vanishing point (Vy) forming an intersection between the projection in the image plane of the first parallel line and the second parallel line to the reference line of the road marker.,

4. A method according to any preceding claim, wherein the score depends on a sum of terms, each term being proportional to the cube of one of the distances.

5. Method according to one of the preceding claims, in which at at least one of the first parametric curve and the second parametric curve is a B-spline, preferably a B-spline of order 3.

6. Method according to one of the preceding claims, comprising a step of polynomial smoothing (108) of at least one of the first parametric curve and the second parametric curve, implemented before the calculation of the road model (110).

7. Method according to one of the preceding claims, in which the camera alignment parameters comprise at least one of the following parameters: • an elevation angle of the camera (2) relative to the road, • a pitch angle of the camera (2) relative to the road, • an azimuth determined from the first value, the second value and the vanishing point of intersection of the tangents (tangent_T 1, tangent_T2), • an image acquisition altitude relative to the road, • a width of the road (L), • a distance (d) on the ground between the road and the camera (2) having acquired the image.

8. Method according to the preceding claim, in which the minimization of the score is carried out on a discrete domain of the first value, the second value, the elevation angle, the pitch angle, the distance (d) on the ground between the road and the camera (2), the acquisition altitude, and the width of the road (L).

9. Computer-readable memory storing computer-executable instructions for carrying out the steps of the method according to one of claims 1 to 8.

10. Road control system (1) comprising: • a camera (2) configured to acquire an image showing a first edge of a road and a second edge of the road opposite the first edge, • a sensor (4), such as a Doppler radar or a lidar, configured to acquire data relating to the trajectory followed by a vehicle traveling on the road, the sensor being fixed relative to the camera (2), • a processing module (6) configured to determine alignment parameters of the camera (2) relative to the road by means of the method according to one of claims 1 to 8.