Method for configuring a road traffic control unit

Abstract

Method for configuring a road control unit comprising the following steps: - the acquisition of a first image of at least one lane of a road by the optical context device and a second image by one or more optical modules, the second image comprising at least a portion of the lane of the first image, the two images being acquired from two different viewing angles and / or focal lengths; - a first selection of at least three points, in the first image, each point; - a second selection of at least three points, in the second image, each point being a counterpart of the three points of the first image; - the determination of a transfer function between the points, of the first image and the points, of the second image, the image having the function of a reference image.

Description

Title of the invention: Method for configuring a road traffic control unit. Technical field

[0001] The present invention relates to a method and a device for configuring a road control unit. Technical background

[0002] It is common practice to control or monitor road traffic using automated systems or control units comprising speedometers—radar or laser devices for measuring the speed and position of vehicles—and optical systems configured for acquiring images of the interior and / or exterior of vehicles. These units are programmed to detect and characterize certain traffic violations committed by vehicle drivers through the combined analysis of speedometer signals and images. In particular, they enable the generation of elements for classifying and proving the violation, and the extraction of identifying information, such as vehicle registration numbers, from the images for issuing a traffic ticket.

[0003] Certain laws, regulations, and / or administrative practices may require so-called contextual information about certain parts of the offending vehicle and / or its surroundings. For example, in Germany, it may be required that the geometric shape delimiting the offending vehicle in the image include a portion of the front of the offending vehicle, the entire license plate of the offending vehicle, and not include any visible elements of another vehicle in the vicinity.

[0004] The reliability of the control units depends in part on the accuracy and robustness of the arrangements and alignments of the speed measuring devices and optical systems with each other in order to ensure, for each vehicle monitored, a perfect match between the data from the speed measuring devices and the images acquired by these systems. This match is essential to eliminate any ambiguity as to whether a vehicle identified as being in violation by the speed measuring devices corresponds exactly to the vehicle whose image is acquired by the optical devices for its identification in accordance with the requirements of the legislation.

[0005] EP 2 690 459 A2, JENOPTIK ROBOT GMBH [DE], 29.01.2014 describes a traffic control system comprising a radar or laser speed gun and two cameras, one of which is a high-resolution camera. The speed gun and the two cameras are arranged so that their respective optical fields overlap. The system includes a data processing unit configured to identify objects in images acquired by the two cameras and compare their positions with those measured by the speedometer.

[0006] It is common practice to manually adjust the positions and orientations of optical devices and speedometers at the factory. These so-called coupling or harmonization adjustments, which may, in some cases, be supplemented by adjustments on the installation site, are based on the ability of a human operator to analyze a number of images acquired by the optical systems during the passage of vehicles and to establish an optimal configuration for all the optical elements in order to ensure both maximum lane coverage and the best possible match between the speedometer data and the acquired images.

[0007] An example of adjusting the coupling between optical devices could consist of determining the focal lengths of the optical lenses and establishing the translation operations and Euler angles involved in the changes of reference frames to perform the projections onto the image planes of the cameras. A transfer function allowing the conversion from the frame of reference of the speedometer to the frame of reference of the optical devices can then be obtained.

[0008] EP 3 012 652 A1, MORPHO [FR], 27.04.2016 describes a road traffic control unit comprising a speed camera and optical devices mechanically assembled using a mounting plate. The mounting plate comprises an assembly of one main part and two secondary parts, each of which is adapted to support an optical device. The three parts include removable fasteners configured to orient the optical devices at a predetermined constant angular orientation relative to a radio axis of the speed camera. Adjustment margins for the orientation are provided to account for possible variations in the widths of the road lanes, the average speed of vehicles on the road, or the topography of the location where the road traffic control unit is installed. The orientation and its adjustment margins are determined by the mounting plate itself for each optical device.This unit simplifies its installation and improves the accuracy of its adjustment to ensure the match between the vehicle monitored by the speed camera and the one appearing in the images acquired by the optical devices.

[0009] WO 2022 / 223921 A1, IDEMIA IDENTITY & SECURITY FRANCE [FR], 27.10.2022 describes an image acquisition device comprising a single-piece stage and three optical devices whose optical axes are coplanar. The optical axes of the three optical devices are oriented such that their field angles partially overlap, and the optical vectors supporting their optical axes satisfy specific vector relations. The stage may include In addition, a fourth optical device is used, the optical axis of which intersects the optical axes of two of the other optical devices at a given point in space. This device allows the acquisition of a low-resolution wide-angle image into which a high-resolution narrow-field image assembly is superimposed. The device is more economical because it eliminates the need for a separate high-resolution wide-angle optical device. Summary of the invention

[0010] One drawback of current control units is their lack of modularity at their operating site, that is, the ability to add, replace, or remove at least one optical device at their operating site, particularly via a module or block that includes said optical device. Each time an optical device is added, replaced, or removed, it is necessary to perform new coupling or harmonization adjustments; an operation that is very often difficult to carry out at the operating or installation site of the control units.

[0011] There is therefore a need for a method allowing modularity of the control units while facilitating their adjustment

[0012] According to a first aspect of the invention, a method for configuring a road traffic control unit is provided, in which: - the road control unit is directed towards an infraction line of at least one traffic lane of a road; - the road control unit includes, an optical context device, at least one optical module, and a controller; said process comprising: - the acquisition of a first image of at least one lane of a road by the optical context device and a second image by one or more optical modules, the second image including at least a portion of the lane of the first image, the two images being acquired from two different viewing angles and / or focal lengths; - a first selection of at least three points, in the first image; - a second selection of at least three points, in the second image, each point being a counterpart of the three points in the first image; - the determination of a transfer function f between the points of the first image and the points of the second image, the image having the function of reference image.

[0013] According to a first embodiment, the transfer function ftrans is a homographic function between the points of the first image and the points of the second image, the first image having the function of a reference image based on the distance to measurement plan of the cinemometer and geometric models of the optical context device and optical modules.

[0014] According to a second embodiment, in the first selection step, each point of the first image is selected near an infringement line, and the transfer function ft is a composite function f. = of three sub- functions fl, f2, f3 in which: - the first sub-function fl is a function of change of coordinate system from the two-dimensional coordinate system of the first image to the three-dimensional coordinate system of the context optical device; - the second sub-function f2 is a change-of-reference function from the three-dimensional reference frame of the context optical device to the three-dimensional reference frame of the optical module, said sub-function f2 comprising at least three adjustment parameters corresponding respectively to the precession, nutation and proper rotation angles of said optical module; - the third sub-function f3 is a change of reference frame function from the three-dimensional frame of the optical module to the two-dimensional frame of the second image; and the step of determining the composite transfer function 7trans = A° A includes a sub-step of optimizing the three parameters adjustment of the sub-function f2 using an optimization algorithm configured to minimize the Euclidean distances between the corresponding points of the second image and the image points of the points in the second image by application of the composite transfer function ft.

[0015] According to a third embodiment, during the first selection step, each point of the first image is selected near an infraction line and contained within a plane previously defined in a three-dimensional coordinate system of the road lane, and the transfer function is a composite function f trans= A° A° A° A four sub-functions f4, f5, f6, f7 in which: - the first sub-function f4 is a change-of-reference function from the two-dimensional reference frame of the first image to the three-dimensional reference frame of the context optical device; - the second sub-function f5 is a change of reference frame function from the three-dimensional frame of the optical context device to the three-dimensional frame of the road lane; - the third sub-function f6 is a coordinate system change function from the three-dimensional coordinate system of the road lane to the three-dimensional coordinate system 903 of the optical module, said sub-function f6 comprising at least three parameters adjustment corresponding respectively to the angles of precession, nutation and proper rotation of said optical module in the three-dimensional frame of reference of the road channel; - the fourth sub-function f7 is a change of reference frame function from the three-dimensional frame of the optical module to the two-dimensional frame of the second image; - the step of determining the composite function f, of transfer f trans= A° A° A ° A includes a substep of optimization of the three adjustment parameters of the subfunction f6 using an optimization algorithm configured to minimize the Euclidean distances between the corresponding points of the second image and the image points of the points in the second image by application of the composite transfer function f.

[0016] According to certain variants of the third embodiment, each point of the first image is considered in the focal plane of the optical context device, and the plane in the three-dimensional frame of the road lane is a plane orthogonal to one of the axes of said three-dimensional frame.

[0017] According to some embodiments, the points of the first image and the second image are selected on either side of the line of violation, and preferably distributed across the width of the traffic lane.

[0018] According to some embodiments, before the optimization step, a step of determining ranges of values ​​for one or more adjustment parameters.

[0019] According to certain embodiments, the points of the first image and the points of the second image are characteristic points of the road lane and / or characteristic points of the rear or front face of a vehicle

[0020] According to some embodiments, the points of the first image and the points of the second image are selected by means of an image processing algorithm adapted to reveal corresponding features between the first image and the second image.

[0021] According to some embodiments, the points of the first image and the points of the second image are selected manually via a user interface communicating with the controller.

[0022] According to some embodiments, the number of points selected on the first image and the second image is between 3 and 50.

[0023] According to certain embodiments, the unit further comprises a speedometer, and a step for applying a correction function f to the r rr corr J trans J corr transfer function f, said correction function f is a function of J traits ' J corr correction between the optical context device and the speedometer and is preceded determined during a prior coupling adjustment between the optical context device and the speedometer.

[0024] According to a second aspect of the invention, a data processing device is provided comprising means for implementing the steps of selecting points in images and the step of determining the composite function f of the process according to any embodiment of a configuration process of the first aspect of the invention.

[0025] According to a third aspect of the invention, a method for installing a road traffic control unit is provided in which: - the road control unit is directed towards an infraction line of at least one traffic lane of a road; - the road control unit includes a speed camera, an optical context device, and a controller; - the optical device and the speedometer having previously been adjusted for coupling according to coupling parameters stored in a memory of the controller; - the coupling parameters include the azimuth of the speedometer in the image plane of the optical context device; The process includes the following steps: - the positioning of a control unit near a road where at least one lane needs to be monitored; - the orientation of the road control unit so that the radio axis of the speed camera and the optical axis of the optical context device are oriented towards at least one lane, in particular towards the line of offence of said lane; - the installation and orientation of at least one optical module so that its optical axis is oriented towards said lane, in particular towards the line of offence, and its optical field is wholly or partly covered by the optical field of the context optical device; - the configuration of the control unit using a method according to any embodiment of a configuration method of the first aspect of the invention.

[0026] According to a fourth aspect of the invention, a method for controlling a vehicle on a road lane is provided, the method comprises the following steps: - the installation of a road control unit by means of an installation method according to the third aspect of the invention; - measuring the speed of at least one vehicle using the speed camera of the road control unit; - the simultaneous or successive acquisition of a first context image of the vehicle using the optical context device and a second image of the registration plate of said vehicle using an optical module, when the speed of said vehicle is greater than a predefined threshold value and / or when said vehicle (103) crosses a line of lights. Brief description of the drawings

[0027] [Fig.1] is a schematic representation of a road controlled by means of a road control unit.

[0028] [Fig.2] is a schematic representation of the structure of a road traffic control unit.

[0029] [Fig.3] is a schematic representation of an example of an image acquisition device comprising a mounting plate, an optical context device and two optical modules.

[0030] [Fig.4] is a process diagram of a method for configuring a road control unit according to the invention.

[0031] [Fig.5] is a schematic representation of a first image IM1 acquired by a context optical device and a second image IM2 acquired by an optical module.

[0032] [Fig.6] is a schematic representation of the image foci and optical fields of a context optical device and two optical modules according to a first configuration.

[0033] [Fig.7] is a schematic representation of the image foci and optical fields of a context optical device and two optical modules according to a second configuration.

[0034] [Fig.8] is a schematic representation of a first image IM1 acquired by a context optical device and a second image IM2 acquired by an optical module.

[0035] [Fig.9] is a schematic representation of the two-dimensional and three-dimensional reference frames of a context optical device and an optical module according to one embodiment.

[0036] [Fig. 10] is a schematic representation of a three-dimensional coordinate system of a road lane, and of the two-dimensional and three-dimensional coordinate systems of a context optical device and an optical module according to an embodiment

[0037] [Fig. 11] is a schematic representation of a data processing device. Detailed description of embodiments

[0038] In the context of the present invention, the term "optical device" means a device capable of acquiring images, photographs or videos of roadways. at given focal lengths and magnifications. Such a device typically includes an optical lens and a photosensitive sensor. The optical device may further include an electronic connector, such as a pin connector, for its power supply and the transmission of acquired images to a data processing device such as a computer.

[0039] The term "photosensitive sensor" means a photosensitive electronic component capable of converting electromagnetic radiation, in particular ultraviolet, visible, or infrared electromagnetic radiation, into an electrical signal, and then processing it to form an image. Such an electronic component is generally devoid of any optical element, such as a lens, capable of forming an optical system whose main function is to modify the path of the electromagnetic radiation before its capture by the active surface of said component. Optionally, in the case of a digital photosensitive sensor, the active surface of the photosensitive sensor may be equipped with microlenses whose function is to concentrate the electromagnetic radiation onto each pixel of the active surface. These microlenses are part of the active surface of the sensor and are not added elements.

[0040] The term "optical objective" means an optical system whose principal function is to modify the path of electromagnetic radiation. In particular, an optical objective is understood to be an optical device formed by a succession of spherical, aspherical, or plane diopters, such as optical lenses, and capable of forming the image of an object located at certain depths of field. A photographic lens is an example of an optical objective that can be used in the context of the invention.

[0041] The term "mounting plate" means any support capable of supporting optical devices and ensuring a secure mechanical connection between them. This support is generally a flat surface with at least one main portion to which the context camera and the optical device(s) can be attached. The support is preferably made of metal or a metal alloy. It may be made of steel or aluminum alloy.

[0042] The term “geometric model” (“camera model”) of an optical device or optical module means a projection model that allows the conversion from the position of points in a three-dimensional scene to their projection onto a two-dimensional image plane of the optical device or optical module. The construction of the geometric model is based on a projection model and the estimation of the intrinsic and extrinsic parameters of the optical device or optical module during a calibration process.

[0043] With reference to [Fig. 1], as an example of a road environment 100, a road control unit 101 is positioned at the edge of a road 102 on which a Vehicle 103 equipped with a license plate 103a. Route 102 can be any type of traffic space authorizing vehicle traffic, for example, a highway, a street, a road, etc. Route 102 comprises two traffic lanes 102a, 102b, delimited by various markings 105 applied to the surface of Route 102 and / or separation elements 106 such as a median strip. The markings 105 generally take the form of visual signs such as a solid line, a broken line, or cones.

[0044] The traffic control unit 101 is generally fixed relative to the traffic lanes 102a and 102b of road 102. It can be mounted on a gantry 104, or, equivalently, on a bridge or gantry spanning the road. It is located at a defined distance from the edge of the road to allow sufficient clearance. It is preferably positioned at a height greater than 0.70 m, or even greater than 2 m, or even greater than 3 m. The elevated positioning of the traffic control unit 101 helps to limit the obstruction of the detection fields of its speedometer and optical systems.

[0045] Generally, the control unit 1001 is oriented towards an offense line 107 which serves as a reference line for speed control. This offense line 107 is generally a virtual line whose position is determined during the installation of the control unit 101. In certain use cases, it may correspond to a traffic light line or a stop line.

[0046] With reference to [Fig. 2], by way of example, a traffic control unit 200 comprises a speed camera 201, a context optical device 202, at least two other optical devices 203, 204 mounted on a plate 205 and whose optical axes are oriented in different directions, a controller 206 equipped with a memory, and a protective housing 207. The radio axis of the speed camera 201 is oriented towards the violation line 107 so that its measuring field covers at least one lane 102a of the road 1002. The context optical device 202 and the two other optical devices 203, 204 are oriented so that their optical fields cover any relevant part of the road 102 at the level of the violation line 107.

[0047] In the context of the invention, the optical devices 203, 204 other than the context optical device 202, are in the form of removable modules or blocks, and are hereinafter referred to as optical modules.

[0048] With reference to [Fig. 3], according to certain embodiments, the context optical device 202 and the optical modules 203, 204 are fixed to a mounting plate 205 which serves as a support. The mounting plate 205 includes locations and fixing means adapted for mechanically securing the context optical device 202 and the optical modules 203, 204. The device optical 202 of context and optical modules 203, 204 can be fixed to it using any suitable fixing means 301 such as, for example, a screw / nut assembly.

[0049] The optical context device 202 is generally fixed at the factory in a non-removable manner, whereas the optical modules 203, 204 are fixed in a removable manner during the installation of the road control unit 101. The optical modules 203, 204 are positioned according to a pre-established configuration, for example using a nomogram, so that their optical axes (AO1), (AO2) are respectively oriented in two directions at different angles α, [3] with respect to the optical axis (AO0) of the context camera 202.

[0050] The context optical device 202 is designed to provide a wide view, or panorama, of an area of ​​the road 102 to be monitored. With reference to [Fig. 6] & [Fig. 7], its optical field c-202 is wider than the optical fields c-203, c-204 of the other optical devices 203, 204. It includes them partially or totally. The context optical device 202 thus makes it possible to capture at least one image that includes the vehicle 103 and its surroundings, for example, a traffic light or the roadside, while the two other optical devices 203, 204 make it possible to capture at least one image of a part of the vehicle 203, in particular a part including its license plate 103a.

[0051] The context optical device 202 may be a wide-angle camera, with high or low resolution. It may include a short focal length lens with a field of view of at least 50°, preferably at least 70°, or even at least 90°. It is preferably a high-resolution camera, enabling the acquisition of images with a resolution of at least 720p, or even 1080p. According to one application example, in the case of traffic light enforcement, the optical field of the context optical device 202 must be wide enough so that a traffic light is visible in the image acquired by said device. According to another application example, the optical field must be such that all traffic lanes 102a, 102b are visible in the acquired image.

[0052] According to some embodiments, the context optical device 202 and the optical modules 203, 204 are monocular optical devices and their projection model is a pinhole model.

[0053] The traffic control unit 200 may include one or more optical modules 203, 204 depending on the number of lanes 102a, 102b of route 102 and / or other prerequisites such as the resolution required for images of license plates and / or the interior of vehicle passenger compartments. For example, some legislation requires a single image of the vehicle (in the United States, in New York for example), while others require several images representing different elements of the vehicle (license plate, vehicle as a whole...) or of its environment (condition of the traffic light,...).

[0054] The controller 206 includes one or more processors for processing the data and signals necessary for the operation of the control unit 200. It further includes data storage memory, for example, of any type such as Flash, EEPROM, HDD, SSD, etc. It may also be connected to a user interface (not shown) to facilitate the configuration of the road control unit 200 by an operator. It may also be connected to a communication device (not shown) to transmit data collected during the monitoring of road 102 to a remote server.

[0055] According to a first aspect of the invention, with reference to [Fig.4] & 5, a method 400 is provided for configuring a road control unit 101, 200, in which: - the road control unit 101, 200 is oriented towards an infringement line 107 of at least one traffic lane 102a, 102b of a road 102; - the road control unit 101, 200 includes, an optical context device 202, at least one optical module 203, 204, and a controller 206; said process comprising: - the acquisition 401 of a first image 501 of at least one lane 102a, 102b of a road 102 by the optical context device 202 and of a second image 502 by one or more optical modules 203, 204, the second image 502 comprising at least a portion of the lane 102a, 102b of the first image 501, the two images 501, 502 being acquired according to two different viewing angles and / or focal lengths; - a first selection 402 of at least three points 501a, 501b, 501c, in the first image 501; - a second selection 403 of at least three points 502a, 502b, 502c, in the second image 502, each point 502a, 502b, 502c being a counterpart of the three points 501a, 501b, 501c of the first image 501; - the determination 404 of a transfer function between the points 501a, 501b, 501c, of the first image 501 and the points 502a, 502b, 502c, of the second image 502, the image 501 having the function of reference image.

[0056] The objective of the configuration is to optically couple the optical modules 203, 204 with the context optical device 202 and the speedometer 201. Generally, before the installation of the control unit 101, 200, only the context optical device 202 and the speedometer 201 have undergone coupling adjustment using a suitable workshop method. During this preliminary workshop adjustment, a correction function between the context optical device 202 and the speedometer 201 can also be determined. Then, using the configuration method according to the invention, each optical module 203, 204 is subject to alignment with the context optical device 202, the context optical device 202 then having a function of alignment reference.

[0057] According to a first embodiment, the transfer function f is a homographic function between the points 501a, 501b, 501c of the first image 501 and the points 502a, 502b, 502c of the second image 502, the first image 501 having the function of reference image from the distance to the measurement plane of the cinemometer 201 and the geometric models of the optical context device 202 and the optical modules 203, 204.

[0058] Homography in computer vision is a linear transformation between two projection planes. A homographic function, generally represented as a matrix, allows us to link two images, here 501 and 502, of the same plane surface in space, here the portion of lane 102a, 102b of a road 102. Calculating a homographic function is a common operation in computer vision. A detailed example is available in Hartley, Richard, and Andrew Zisserman. Multiple view geometry in computer vision. Cambridge University Press, 2003.

[0059] The distance and orientation of the optical modules 203, 204 relative to the context optical device 202 determine the position of their image foci. It is then possible to distinguish a first configuration in which the foci can be considered as coinciding and a second configuration in which the foci are too far apart to allow such an approximation.

[0060] According to the first configuration, with reference to [Fig. 6], when the distances between the optical modules 203, 204 and the context optical device 2002 are small, typically on the order of 50 mm, the image foci f-202, f-203, f-2004 of the optical modules 203, 204 and the context optical device 202 can be considered to coincide at a single point. The optical fields c-203 and c-204 of the optical modules 203, 204 are then generally covered by the optical field c-202 of the context optical device 202.

[0061] According to the second configuration, with reference to [Fig. 7], when the distances between the optical modules 203, 204 and the context optical device 202 are significant, typically greater than 50 mm, the image foci f-202, f-203, f-204 of the optical modules 203, 204 and the context optical device 202 cannot be considered to coincide. The optical field c-202 of the context optical device 202 then only partially covers the optical field c-203 of at least one of the two optical modules 202, 203.

[0062] According to a second embodiment, with reference to [Fig. 8] & 9, during the first selection step 402, each point 801a, 801b, 801c of the first image 801 is selected near an infringement line 107, and the transfer function is a composite function of three sub-functions f1, f2, f3 in which: - the first sub-function fl is a change of reference frame function from the two-dimensional frame 901 of the first image 801 to the three-dimensional frame 902 of the context optical device 202; - the second sub-function f2 is a change-of-reference function from the three-dimensional reference frame 902 of the context optical device 202 to the three-dimensional reference frame 903 of the optical module 203, 204, said sub-function f2 comprising at least three adjustment parameters corresponding respectively to the angles of precession, nutation and proper rotation of said optical module 203, 204; - the third sub-function f3 is a change of reference frame function from the three-dimensional frame 903 of the optical module 203, 204 to the two-dimensional frame 904 of the second image 802; and the step 404 for determining the composite transfer function ^tram = f ° ^2 A includes a substep of optimization of the three adjustment parameters of the subfunction f2 using an optimization algorithm configured to minimize the Euclidean distances between the corresponding points 802a, 802b, 802c of the second image 802 and the image points 801a-1, 801b-2, 801c-3 of the points 801a, 801b, 801c in the second image 802 by application of the composite transfer function f. ■' tram

[0063] This second embodiment is particularly advantageous when the optical modules 2003, 2004 and the optical context device 2002 are arranged according to the first configuration, i.e., with reference to [Fig.6], so that their image foci f-2002, f-2003, f-2004 can be considered as coinciding.

[0064] In [Fig.9], for the sake of clarity of illustration, only one optical module 203, 204 is shown, and the distance scales as well as the relative positions of the context optical device 202 and the optical module 203 are intentionally not respected.

[0065] In the example shown in [Fig. 8], the image points 801a-1, 801b-2, 801c-3 are represented in the second image 802 by empty circles. The corresponding points 802a, 802b, 802c are represented in the second image 802 by filled circles. The image points 801a-1, 801b-2, 801c-3 are the images of 801a, 801b, 801c in image 802 by application of the intermediate sub-function f2.

[0066] Preferably, in practice, the points 801a, 802b, 802c selected in the first image 801 are sufficiently far from the context optical device 202 that they can be considered at infinity with respect to the center of the three-dimensional frame 902 of said context optical device 202. In other words, the value of the Z-coordinate of the points 801a, 802b, 802c in the three-dimensional frame 902 of the context optical device 202 is large enough that these points can be considered as an approximation of a position located at infinity with respect to the center of the three-dimensional frame 902 of said context optical device 202.

[0067] Such an approximation advantageously ensures that the points 801a, 801b, 801c selected in the first image 801 have image points 801a-l, 801b-2, 801c-3 obtained by application of the intermediate sub-function f2 in the second image 802. In other words, the approximation prevents the image points 801a-l, 801b-2, 801c-3 obtained by application of the composite transfer function ft from being outside the field of the optical module 203, 204, i.e. outside the second image 802.

[0068] In [Fig. 8], the violation line 107 is represented as a virtual line serving as a reference line for speed control by the control unit 101, 200. In other examples, the violation line may be a traffic light line or a stop line. Preferably, points 801a, 801b, 801c of the first image 801 and 802a, 802b, 802c of the second image 802 are selected on either side of the violation line 107, and preferably, distributed across the width of the traffic lane 102a, 102b.

[0069] The first sub-function fl of the change of reference frame and the third sub-function f3 of the change of reference frame are generally geometric models ("camera model"). They allow the conversion from the position of points in a three-dimensional scene to their projection onto a two-dimensional image plane and vice versa. The construction of the geometric model is based on a projection model, for example a stenographic model, and the estimation of the intrinsic and extrinsic parameters of the optical device or optical module during a calibration process.

[0070] According to some variants, the optimization algorithm is a Gaussian reduction method, that is, a minimization of the quadratic sum of the Euclidean distances. The Euclidean distances may optionally be assigned a weighting factor corresponding to each of the points.

[0071] According to some variations, the method includes, before the optimization step, a step of determining ranges of values ​​for one or more parameters adjustment of the second sub-function f2. These value intervals can be derived from topographic data (distance from road edge 102, height of control unit 101, number of lanes 102a, 102b) and the orientation of control unit 101 (left or right sighting). According to some complementary variants, the optimization algorithm can be a local gradient descent algorithm by exploring the values ​​of the previously determined intervals, or even metaheuristic algorithms for finding a global minimum.

[0072] Regardless of the embodiment, according to certain variants, with reference to [Fig. 5] & 8, the points (501a, 501b, 501c), (801a, 801b, 801c) of the first image 501, 801 and the points (502a, 502b, 502c), (802a, 802b, 802c) of the second image 502, 802 are characteristic points of lane 102a, 102b of route 102. In the example illustrated in [Fig. 5] & 8, the characteristic points are points of the road marking 105: the points 501a, 501b, 502a, 502b, 801a, 801b, 802a, 802b are points of the lateral broken line 105a of Route 102; points 501c, 502c, 801c, 802c, are points on the center broken line 105b of Route 102. Depending on the location of the 101, 200 traffic control unit and the type of any 102, for example a line of lights at an intersection, the characteristic points may be points on the direction arrows, pedestrian crossings, or the lines of lights.In the absence of road markings, the characteristic points may be on a sign element, such as a road sign, traffic lights, or curbs, located in the immediate vicinity of the road.

[0073] Alternatively or in addition, the points (501a, 501b, 501c), (801a, 801b, 801c) of the first image 501, 801 and the points (502a, 502b, 502c), (802a, 802b, 802c) of the second image 502, 802 may be characteristic points of the rear or front face of a vehicle 103, in particular points located in the registration plate 103a.

[0074] According to some variants, the points (501a, 501b, 501c), (801a, 801b, 801c) of the first image 501, 801 and the points (502a, 502b, 502c), (802a, 802b, 802c) of the second image 502, 802 are selected using an image processing algorithm adapted to reveal corresponding features between the first image 501, 801 and the second image 502, 802. Examples of algorithms are descriptor / detector type algorithms such as SIFT, SURF, BRIEF, and ORB, or neural network type algorithms, such as SIAMESE neural networks. The following article may be advantageously consulted: Ma et al. (2021). Image matching from handcrafted to deep features: A survey. International Journal of Computer Vision, 129(1), 23-79; Karami et al. (2017) Image matching using SIFT, SURF, BRIEF and ORB: performance comparison for distorted images. arXiv preprint arXiv: 1710.02726 ; Hanif, MS (2019) Patch match networks: Improved two-channel and Siamese networks for image patch matching. Pattern Recognition Letters, 120, 54-61.

[0075] According to some alternative variants, the points (501a, 501b, 501c), (801a, 801b, 801c) of the first image 501, 801 and the points (502a, 502b, 502c), (802a, 802b, 802c) of the second image 502, 802 are selected manually via a user interface communicating with the controller 205. These variants are advantageous when the road control unit 101 is installed near roads 102 lacking characteristic features such as road markings or signaling devices. This type of situation can be encountered, in particular, in non-urbanized areas, for example, in the countryside.

[0076] According to some variants, the number of points selected on the first image 501, 801 and the second image 502, 802 is between 3 and 50. The greater the number of points, the more precise the coupling between the speedometer 201 and the optical modules 203, 204 will be, in particular for the method according to the first embodiment.

[0077] According to a third embodiment, with reference to [Fig.8] & 10, during the first selection step 402, each point 801a, 801b, 801c of the first image 801 is selected near an infringement line 107 and contained in a plane 1001 previously defined in a three-dimensional frame 1002 of the lane 102a, 102b of the road 102, and the transfer function, is a composite function ftrans = / 4° f5° f6° fj of four sub-functions f4, f5, f6, f7 in which: - the first sub-function f4 is a change-of-frame function from the two-dimensional frame 901 of the first image 801 to the three-dimensional frame 902 of the context optical device 202; - the second sub-function f5 is a change of reference frame function from the three-dimensional frame 902 of the optical context device 202 to the three-dimensional frame 1002 of the track 102a, 102b of the road 102; - the third sub-function f6 is a change-of-reference function from the three-dimensional reference frame 1002 of lane 102a, 102b of road 102 to the three-dimensional reference frame 903 of the optical module 203, 204, said sub-function f6 comprising at least three adjustment parameters corresponding respectively to the angles of precession, nutation and proper rotation of said optical module 203, 204 in the three-dimensional reference frame 1002 of lane 102a, 102b of road 102; - the fourth sub-function f7 is a change of reference frame function from the three-dimensional frame 903 of the optical module 203, 204 to the two-dimensional frame 904 of the second image 802; - The step 404 of determining the composite transfer function ft f trans ~ A° A ° A includes a sub-step of optimizing the three parameters adjustment of the sub-function f6 using an optimization algorithm configured to minimize the Euclidean distances between the corresponding points 802a, 802b, 802c of the second image 802 and the image points 801a-1, 801b-2, 801c-3 of the points 801a, 801b, 801c in the second image 802 by applying the composite transfer function / . J traits

[0078] This third embodiment is particularly advantageous when the optical modules 203, 204 and the optical context device 202 are arranged according to the second configuration, i.e., with reference to [Fig.7], so that their image foci f-2002, f-2003, f-2004 cannot be considered to coincide.

[0079] In [Fig. 10], for the sake of clarity, only one optical module 203, 204 is shown, and the distance scales as well as the relative positions of the contextual optical device 202 and the optical module 203, 204 are intentionally omitted. For the same reasons, only a point 801a of the first image 801, a point 802a of the second image 801, and an image point 801a-1 of point 801a in the second image 802 are shown.

[0080] In the example illustrated in [Fig. 10], the point 801a of the first image 801 corresponds to the point 1003 contained in the plane 1001 in the three-dimensional frame 1002 of the lane 102a of the road 102. This correspondence is illustrated by a straight line 1004 which passes through the center of the three-dimensional frame 902 of the optical device 202 of context and the point 801a of the first image 801.

[0081] The passage from point 801a of the first image 801 to point 1003 contained in plane 1001 in the three-dimensional frame 1002 of lane 102a of road 102 is obtained in the following manner: 1 - Apply the sub-function f4 to the point 801a of the first image 801 to obtain an image point f4(801a) of said point 801a in the three-dimensional frame 902 of the context optical device 202 considering the Z coordinate of the image point f4(801a) equal to the focal length of the context optical device 202; 2 - Apply the sub-function f5 on the image point f4(801a) obtained and on the origin of the three-dimensional frame 902 of the optical device 202 of context, to obtain the points f5(f4(801a) and f5(902) respectively; 3 - Calculation of the coordinates of the point of intersection 1003 in the three-dimensional frame 1002 of the track 102a, 102b between a line passing through the points f5(f4(801a) and f5(902) with the plane 1001 defined in the three-dimensional frame 1002 of the track 102a, 102b - in the example shown in [Fig.10], the plane 1001 is defined as contained in the track 102a, 102b.

[0082] The various aspects detailed in the second embodiment of the process are directly applicable to the first embodiment.

[0083] On [Fig.10], the image point 801a-l of the point 801a in the second image 802 is represented by an empty circle. It corresponds to point 1003 contained in plane 1001 in the three-dimensional frame 1002 of lane 102a of road 102. This correspondence is illustrated by a straight line 1005 that passes through the center of the three-dimensional frame 903 of the optical module 203, 204 and point 1003 contained in plane 1001 in the three-dimensional frame 1002 of lane 102a of road 102. The intersection of this straight line 1005 with the plane of the second image 802 represents the image point 801a-l of point 801a in the second image 802. The passage from point 1003 contained in plane 1001 in the three-dimensional frame 1002 of lane 102a, 102b of road 102 to image point 801a-l is obtained by applying the sub-functions f7.

[0084] In the determination step 404, the parameters of sub-function f6 are adjusted so as to minimize the Euclidean distance between the corresponding point 802a, of the second image 802 and the image point 801a-l of the point 801a in the second image 802. This operation is carried out for each of the points 802a, 802b, 802c and the image points 801a-l, 801b-2 and 801c-3.

[0085] Preferably, each point 801a, 801b, 801c of the first image 801 is considered in the focal plane of the context optical device 202, and the plane 1001 in the three-dimensional frame 1002 of the road lane 102 is a plane orthogonal to one of the x, y, z axes of said three-dimensional frame 1002.

[0086] According to some embodiments, the unit 101, 200 further comprises a speedometer 201, and the method 400 further comprises a step of applying / ft a correction function f to the transfer function C, J corr J tram J corr J tram , said function / correction is a correction function between the context optical device 202 and the speedometer 201 and is predetermined during a prior coupling adjustment between the context optical device 202 and the speedometer 201.

[0087] As explained previously, the optical device 202 and the speedometer 201 can undergo a preliminary coupling adjustment in the workshop, and a correction function fcorr can be determined. This preliminary adjustment may include, in particular, the measurement of the azimuth in the horizontal coordinate system (alt-azimuth coordinate system) of the road control unit 101, 200. By way of example, an azimuth measurement can be carried out in the workshop using a Doppler simulation device directly above which an optical target is placed. The azimuth corresponds, in the horizontal coordinate system of the unit 101, 200 In road traffic control, the azimuth is the angle between, on the one hand, a reference direction defined by the context optical device 202 and the target – the optical axis of the context optical device 202 serving as the reference direction – and, on the other hand, the direction defined by the speedometer 201 and the Doppler simulation device. In other words, the position of the target in the image plane of the context optical device 202 corresponds to the position of the Doppler simulation device as detected by the speedometer, and the angle between these two positions from the center of the coordinate system centered on the road traffic control unit 101, 200 corresponds to the azimuth. The correction function between the context optical device 202 and the speedometer 201 is an azimuth correction function.

[0088] According to a second aspect of the invention, with reference to [Fig. 11], the steps of selecting points 402, 403 in the images 501, 502, 801, 802 and of determining the transfer function 404 of the process according to the first aspect of the invention can be implemented by a data processing device 1100 comprising means for implementing said steps. Such a device 1100 is responsible for automatically executing sequences of arithmetic or logical operations to perform tasks or actions.

[0089] According to some embodiments, the data processing device 1000 is an integral part of the controller 205 of the road control unit 101, 200. According to other embodiments, the data processing device 1100 is an added device such as a mobile computer. The controller 205 of the road control unit 101, 200 may then include an input / output interface suitable for communication with the added device. Once the configuration is complete, that is, once the added device has finished executing the configuration process according to the first aspect of the invention, the configuration parameters, including the composite function, are stored in a memory of the controller 205 of the road control unit 101, 200 for subsequent use.

[0090] Whether the data processing device 1100 is an integral part of the traffic control unit 101 controller 205 or a separate device, it includes one or more central processing units (CPUs) 1101 and / or one or more graphics processing units (GPUs) 1102, a physical communication module 1103, one or more physical input / output modules 1104 for data exchange with external devices, a transient storage medium 1105 such as random access memory (RAM), a non-transient recording medium 1106, and communication buses (not shown) for data transfer between the internal components of the data processing device 1100.

[0091] The data processing device 1100 allows the execution of one or more program modules comprising instructions which, when the module(s) The program modules are executed, causing the data processing device 1100 to implement the method according to the invention. The program module(s) can be written in any programming language, compiled or interpreted. They can be part of a software solution, that is, a collection of executable instructions, code, scripts, or other elements, and / or databases.

[0092] According to a third aspect of the invention, a method is provided for installing a 101, 200 road traffic control unit in which: - the road control unit 101 is directed towards an offence line 105 of at least one lane 102a, 102b of traffic of a road 102; - Unit 101, 200 of road control includes a speed camera 201, an optical context device 202, and a controller 205; - the optical device 202 and the speedometer 201 having previously been subjected to a coupling adjustment according to coupling parameters stored in a memory of the controller 205; - the coupling parameters include the azimuth of the cinemometer 201 in the image plane of the context optical device 202; The process includes the following steps: - the positioning of a 101, 200 control unit near a 102 road where at least one lane 102a, 102b is to be monitored; - the orientation of the road control unit 101, 200 so that the radio-electric axis of the speed camera 201 and the optical axis of the optical context device 202 are oriented towards at least one lane 102a, 102b, in particular towards the offence line 107 of said lane 102a, 102b; - the installation and orientation of at least one optical module 203, 204 so that its optical axis is oriented towards said lane 102a, 102b, in particular towards the infringement line 107, and its optical field is wholly or partly covered by the optical field of context optical device 202; - the configuration of the control unit 101, 200 using a method according to the first aspect of the invention.

[0093] The installation and orientation step of at least one optical module 203, 204 can be implemented by an operator using a reticle based on the topographic characteristics of the installation site.

[0094] According to a fourth aspect of the invention, a method is provided for controlling a vehicle 103 on a lane 102a, 102b of a road 102; the method comprises the following steps: - the installation of a 101, 200 road control unit by means of an installation method according to the second aspect of the invention; - the measurement of the speed of at least one vehicle 103 using the speed camera 201 of the road control unit 101, 200; - the simultaneous or successive acquisition of a first context image of the vehicle 103 using the context optical device 202 and a second image of the registration plate of said vehicle 103 using an optical module 203, 204, when the speed of said vehicle 103 is greater than a predefined threshold value and / or when said vehicle 103 crosses a line of lights. References Literature patent

[0095] EP 2 690 459 A2, JENOPTIK ROBOT GMBH [DE], 01.29.2014.

[0096] EP 3 012 652 Al, MORPHO [FR], 04.27.2016.

[0097] WO 2022 / 223921 A1, IDEMIA IDENTITY & SECURITY FRANCE [FR], 27.10.2022. Littérature non-brevet

[0098] Hartley, Richard, and Andrew Zisserman. Multiple view geometry in computer vision. Cambridge university press, 2003.

[0099] Ma et al. (2021). Image matching from handcrafted to deep features: A survey. International Journal of Computer Vision, 129(1), 23-79.

[0100] Karami et al. (2017) Image matching using SIFT, SURF, BRIEF and ORB: performance comparison for distorted images. arXiv preprint arXiv: 1710.02726.

[0101] Hanif, M. S. 2019. Patch match networks: Improved two-channel and Siamese networks for image patch matching. Pattern Récognition Letters, 120, 54-61.

Claims

Demands

1. Method (400) of configuring a road control unit (101, 200), wherein: - the road control unit (101, 200) is oriented towards an infringement line (107) of at least one traffic lane (102a, 102b) of a road (102); - the road control unit (101, 200) comprises, an optical context device (202), at least one optical module (203, 204), and a controller (206); said method comprising: - the acquisition (401) of a first image (501) of at least one lane (102a, 102b) of a road (102) by the optical context device (202) and of a second image (502) by one or more optical modules (203, 204), the second image (502) comprising at least a portion of the lane (102a, 102a) of the first image (501), the two images (501, 502) being acquired according to two different viewing angles and / or focal lengths; - a first selection (402) of at least three points (501a, 501b, 501c), in the first image (501);- a second selection (403) of at least three points (502a, 502b, 502c) in the second image (502), each point (502a, 502b, 502c) being a counterpart of the three points (501a, 501b, 501c) of the first image (501); - the determination (404) of a transfer function f between the points (501a, 501b, 501c) of the first image (501) and the points (502a, 502b, 502c) of the second image (502), the image (501) having the function of reference image.

2. Method (400) according to claim 1, wherein the transfer function is a homographic function between the points (501a, 501b, 501c) of the first image (501) and the points (502a, 502b, 502c) of the second image (502), the first image (501) having the function of a reference image from the distance to the measurement plane of the speed camera (201) and the geometric models of the optical context device (202) and the optical modules (203, 204).

3. Method (400) according to claim 1, the first selection step (402), each point (801a, 801b, 801c) of the first image

4. (801) is selected near a violation line (107), and the transfer function f is a composite function f. = f ° f* f, of three sub-functions fl, f2, f3 in which: t1 uns 1 zd 1 - the first sub-function fl is a change of reference frame function from the two-dimensional frame (901) of the first image (801) to the three-dimensional frame (902) of the context optical device (202); - the second sub-function f2 is a change-of-reference function from the three-dimensional reference frame (902) of the context optical device (202) to the three-dimensional reference frame (903) of the optical module (203, 204), said sub-function f2 comprising at least three adjustment parameters corresponding respectively to the precession, nutation and proper rotation angles of said optical module (203, 204); - the third sub-function f3 is a change of reference frame function from the three-dimensional frame (903) of the optical module (203, 204) to the two-dimensional frame (904) of the second image (802); and the step (404) of determining the composite transfer function ftrans - f° f?° includes an optimization substep of the three adjustment parameters of the subfunction f2 using an optimization algorithm configured to minimize the Euclidean distances between the corresponding points (802a, 802b, 802c) of the second image (802) and the image points (801a-1, 801b-2, 801c-3) of the points (801a, 801b, 801c) in the second image (802) by application of the composite transfer function. A method (400) according to claim 1, wherein in the first selection step (402), each point (801a, 801b, 801c) of the first image (801) is selected near an infraction line (107) and contained within a plane (1001) previously defined in a three-dimensional coordinate system (1002) of the lane (102a, 102b) of the road (102), and the transfer function f is a composite function of four sub-functions f4, f5, f6, f7, wherein: - The first sub-function f4 is a change-of-coordinate function from the two-dimensional coordinate system (901) of the first image (801) to the three-dimensional frame (902) of the optical context device (202); - the second sub-function f5 is a change-of-frame function from the three-dimensional frame (902) of the optical context device (202) to the three-dimensional frame 1002 of the lane (102a, 102b) of the road (102); - the third sub-function f6 is a change-of-frame function from the three-dimensional frame (1002) of the lane (102a, 102b) of the road (102) to the three-dimensional frame 903 of the optical module (203, 204), said sub-function f6 comprising at least three adjustment parameters corresponding respectively to the precession, nutation and proper rotation angles of said optical module (203, 204) in the three-dimensional frame (1002) of the lane (102a, 102b) of the road (102);- the fourth sub-function f7 is a change-of-reference function from the three-dimensional frame (903) of the optical module (203, 204) to the two-dimensional frame (904) of the second image (802); - the determination step (404) of the composite transfer function ft ftrans = / 4° f5° f6° / 7 includes a sub-step of optimization of the three fitting parameters of the sub-function f6 using an optimization algorithm configured to minimize the Euclidean distances between the corresponding points (802a, 802b, 802c) of the second image (802) and the image points (801a-1, 801b-2, 801c-3) of the points (801a, 801b, 801c) in the second image (802) by application of the composite transfer function f.

5. Method (400) according to claim 4, wherein each point (801a, 801b, 801c) of the first image (801) is considered in the focal plane of the context optical device (202), and the plane (1001) in the three-dimensional frame (1002) of the road lane (102) is a plane orthogonal to one of the axes (x, y, z) of said three-dimensional frame (1002).

6. A method according to any one of claims 3 to 5, wherein points (801a, 801b, 801c) of the first image (801) and (802a, 802b, 802c) of the second image (802) are selected on either side of the offence line (107) and preferably distributed across the width of the traffic lane (102a, 102b).

7. A method according to any one of claims 3 to 6, such that prior to the optimization step, a step of determining ranges of values ​​for one or more adjustment parameters.

8. Method (400) according to any one of claims 1 to 7, wherein the points (501a, 501b, 501c), (801a, 801b, 801c) of the first image (501, 801) and the points (502a, 502b, 502c), (802a, 802b, 802c) of the second image (502, 802) are characteristic points of the lane (102a, 102b) of the road (102) and / or characteristic points of the rear or front face of a vehicle (103).

9. Method (400) according to any one of claims 1 to 8, wherein the points (501a, 501b, 501c), (801a, 801b, 801c) of the first image (501, 801) and the points (502a, 502b, 502c), (802a, 802b, 802c) of the second image (502, 802) are selected by means of an image processing algorithm adapted to reveal corresponding features between the first image (501, 801) and the second image (502, 802).

10. Method (400) according to any one of claims 1 to 8, wherein the points (501a, 501b, 501c), (801a, 801b, 801c) of the first image (501, 801) and the points (502a, 502b, 502c), (802a, 802b, 802c) of the second image (502, 802) are selected manually via a user interface communicating with the controller (206).

11. Method (400) according to any one of claims 1 to 10, wherein the number of points selected on the first image (501, 801) and the second image (502, 802) is between 3 and 50.

12. Method (400) according to any one of claims 1 to 11, wherein the unit (101, 200) further comprises a speedometer (201), and an application step f, of a correction function f to the transfer function f, said correction function f is a correction function between the context optical device (202) and the speedometer (201) and is predetermined during a pre-coupling adjustment between the context optical device (202) and the speedometer (201).

13. Data processing device (1100) comprising means for carrying out the steps of selecting (402, 403) points in images (501, 502, 801, 802) and determining (404) the transfer function of the process (400) according to any one of claims 1 to 12.

14. Method of installing a traffic control unit (101, 200) wherein: - the traffic control unit (101, 200) is oriented towards an infraction line (107) of at least one traffic lane (102a, 102b) of a road (102); - the traffic control unit (101, 202) comprises a speed camera (201), a context optical device (202), and a controller (205); - the optical device (202) and the speed camera (201) having previously been adjusted for coupling according to coupling parameters stored in a memory of the controller (205); - the coupling parameters include the azimuth of the speed camera (201) in the image plane of the context optical device (202); the process includes the following steps: - positioning a control unit (101, 200) near a road (102) of which at least one lane (102a, 102b) is to be monitored;- the orientation of the road traffic control unit (101, 200) so that the radio axis of the speed camera (201) and the optical axis of the optical context device (202) are oriented towards at least one lane (102a, 102b), in particular towards the violation line (107) of said lane (102a, 102b); - the installation and orientation of at least one optical module (203, 204) so ​​that its optical axis is oriented towards said lane (102a, 102b), in particular towards the violation line (107), and its optical field is wholly or partly covered by the optical field of the optical context device (202); - the configuration of the control unit (101, 200) using a method according to any one of claims 1 to 12.

15. A method for controlling a vehicle (103) on a lane (102a, 102b) of a road (102), the method comprises the following steps: - the installation of a road control unit (101, 200) by means of an installation method according to claim 13; - the measurement of the speed of at least one vehicle (103) using the speed camera (201) of the road control unit (101, 200); - the simultaneous or successive acquisition of a first context image of the vehicle (103) using the context optical device (202) and a second image of the registration plate of said vehicle (103) using an optical module (203, 204), when the speed of said vehicle (103) is greater than a predefined threshold value and / or when said vehicle (103) crosses a line of lights.